SMJ

‌E-ISSN 2228-8082

Volume 77, Number 12, December 2025

Siriraj Medical Journal

is deeply sadden by the passing of

Her Majesty Queen Sirikit The Queen Mother

We pay tribute to Her Majesty’s lifelong devotion to the nation and the people of Thailand

Siriraj Medical Journal

The world-leading biomedical science of Thailand

MONTHLY

Indexed by

THAILAND SECTION 1954

https://he02.tci-thaijo.org/index.php/sirirajmedj/index

E-mail: sijournal92@gmail.com

ORIGINAL ARTICLE REVIEW ARTICLE

‌ORIGINAL ARTICLE‌

829 The Efficacy of Deep Learning Model on

SMJ

Siriraj Medical Journal

The world-leading biomedical science of Thailand

Volume 77 Number 12

December 2025

Sex Estimation in Patellae for A Thai Population Saranya Honghimaphan, Patara Rattanachet, Natipong Chatthai, Parawee Jitrabeab, Sittiporn Ruengdit, Robert W.Mann, Napakorn Sangchay

847 In Silico Study of Functional Network Analysis Focusing on Antipsychotic Drugs and Cognitive Function in Schizophrenia

Kamonwan Thanontip, Banthit Chetsawang, Vorasith Siripornpanich, Vasunun Chumchua, Jirapa Chetsawang

858 Sexual Dimorphism in Cranial and Post-Cranial Skeletal Elements: Forensic Implications for Sex Estimation in a Contemporary Thai Population

Napakorn Sangchay, Kavin Tangmanpakdeepong, Saidontree Boonyarud, Sujinthara Wansopha, Natipong Chatthai, Patara Rattanachet, Jirapa Chetsawang

877 Minimum Ischiopubic Ramus Width as a Single Metric for Sex Estimation: A Pilot Study

Robert W. Mann, Sittiporn Ruengdit, Patara Rattanachet, Napakorn Sanchay

886 Calcaneal Articular Talar Facets, Stieda’s Process, and Calcaneus Secundarius: Variations Found in Thai Population

901

From Cadaveric Dissection to Artificial Intelligence: A Chronological Review of Advances in Anatomy Education

Anuch Durongphan

REVIEW ARTICLE

Areeya Marie B. Wongla, Patara Rattanachet, Saranya Honghimaphan, Natipong Chatthai, Parawee Jitrabeab, Napakorn Sangchay

‌Executive Editor:

SIRIRAJ MEDICAL JOURNAL

SMJ

https://he02.tci-thaijo.org/index.php/sirirajmedj/index

Professor Apichat Asavamongkolkul, Mahidol University, Thailand

Editorial Director:

Professor Aasis Unnanuntana, Mahidol University, Thailand

Associate Editors

Assistant Professor Adisorn Ratanayotha, Mahidol University, Thailand Pieter Dijkstra, University of Groningen, Netherlands

Professor Phunchai Charatcharoenwitthaya, Mahidol University, Thailand Professor Varut Lohsiriwat, Mahidol University, Thailand

International Editorial Board Members

Editor-in-Chief:

Professor Thawatchai Akaraviputh, Mahidol University, Thailand

Allen Finley, Delhousie University, Canada

Christopher Khor, Singapore General Hospital, Singapore Ciro Isidoro, University of Novara, Italy

David S. Sheps, University of Florida, USA

David Wayne Ussery, University of Arkansas for Medical Sciences, USA Dennis J. Janisse, Medical College of Wisconsin, USA

Dong-Wan Seo, University of Ulsan College of Medicine, Republic of Korea Folker Meyer, Argonne National Laboratory, USA

Frans Laurens Moll, University Medical Center Ultrecht, Netherlands George S. Baillie, University of Glasgow, United Kingdom

Gustavo Saposnik, Unity Health Toronto, St. Micheal Hospital, Canada Harland Winter, Harvard Medical School, USA

Hidemi Goto, Nagoya University Graduate School of Medicine, Japan Ichizo Nishino, National Institute of Neuroscience NCNP, Japan Intawat Nookaew, University of Arkansas for Medical Sciences, USA James P. Doland, Oregon Health & Science University, USA

John Hunter, Oregon Health & Science University, USA

Karl Thomas Moritz, Swedish University of Agricultural Sciences, Sweden Kazuo Hara, Aichi Cancer Center Hospital, Japan

Keiichi Akita, Institute of Science Toko, Japan Kyoichi Takaori, Kyoto University Hospital, Japan Marcela Hermoso Ramello, University of Chile, Chile Marianne Hokland, University of Aarhus, Denmark

Matthew S. Dunne, Institute of Food, Nutrition, and Health, Switzerland Mazakayu Yamamoto, Tokyo Women’s Medical University, Japan Mitsuhiro Kida, Kitasato University & Hospital, Japan

Moses Rodriguez, Mayo Clinic, USA

Nam H. CHO, Ajou University School of Medicine and Hospital, Republic of Korea Nima Rezaei, Tehran University of Medical Sciences, Iran

Noritaka Isogai, Kinki University, Japan

Philip A. Brunell, State University of New York At Buffalo, USA Philip Board, Australian National University, Australia Ramanuj Dasgupta, Genome Institution of Singapore

Richard J. Deckelbaum, Columbia University, USA Robert W. Mann, University of Hawaii, USA

Robin CN Williamson, Royal Postgraduate Medical School, United Kingdom Sara Schwanke Khilji, Oregon Health & Science University, USA

Seigo Kitano, Oita University, Japan Seiji Okada, Kumamoto University

Shomei Ryozawa, Saitama Medical University, Japan Shuji Shimizu, Kyushu University Hospital, Japan

Stanlay James Rogers, University of California, San Francisco, USA Stephen Dalton, Chinese University of HK & Kyoto University

Tai-Soon Yong, Yonsei University, Republic of Korea Tomohisa Uchida, Oita University, Japan

Victor Manuel Charoenrook de la Fuente, Centro de Oftalmologia Barraquer, Spain Wikrom Karnsakul, Johns Hopkins Children’s Center, USA

Yasushi Sano, Director of Gastrointestinal Center, Japan Yik Ying Teo, National University of Singapore, Singapore Yoshiki Hirooka, Nagoya University Hospital, Japan

Yozo Miyake, Aichi Medical University, Japan Yuji Murata, Aizenbashi Hospital, Japan

Editorial Board Members

Vitoon Chinswangwatanakul, Mahidol University, Thailand Jarupim Soongswang, Mahidol University, Thailand

Jaturat Kanpittaya, Khon Kaen University, Thailand Nopphol Pausawasdi, Mahidol University, Thailand Nopporn Sittisombut, Chiang Mai University, Thailand Pa-thai Yenchitsomanus, Mahidol University, Thailand Pornprom Muangman, Mahidol University, Thailand Prasit Wattanapa, Mahidol University, Thailand

Prasert Auewarakul, Mahidol University, Thailand Somboon Kunathikom, Mahidol University, Thailand Supakorn Rojananin, Mahidol University, Thailand

Suttipong Wacharasindhu, Chulalongkorn University, Thailand Vasant Sumethkul, Mahidol University, Thailand

Watchara Kasinrerk, Chiang Mai University, Thailand Wiroon Laupattrakasem, Khon Kaen University, Thailand Yuen Tanniradorn, Chulalongkorn University, Thailand

Editorial Assistant: Nuchpraweepawn Saleeon, Mahidol University, Thailand

Proofreader: Amornrat Sangkaew, Mahidol University, Thailand, Nuchpraweepawn Saleeon, Mahidol University, Thailand

‌The Efficacy of Deep Learning Model on Sex Estimation in Patellae for A Thai Population

Saranya Honghimaphan, B.Sc.1, Patara Rattanachet, Ph.D.1, Natipong Chatthai, M.Sc.1, Parawee Jitrabeab, M.Sc.1, Sittiporn Ruengdit, Ph.D.2, Robert W.Mann, Ph.D.3, Napakorn Sangchay, M.D., Ph.D.1,*

1Siriraj Anatomical and Anthropological Bone Research Centre (Virapan Davivongs), Department of Anatomy, Faculty of Medicine Siriraj Hospital,

Mahidol University, Bangkok, Thailand, 2Department of Forensic Medicine, Faculty of Medicine, Chiang Mai University, Chiang Mai, Thailand, 3John

A. Burns School of Medicine, University of Hawai‘i, Honolulu, Hawai‘i, USA.

*Corresponding author: Napakorn Sangchay E-mail: napakorn.sac@mahidol.ac.th

Received 13 August 2025 Revised 10 October 2025 Accepted 18 October 2025 ORCID ID:http://orcid.org/0000-0001-9250-3702 https://doi.org/10.33192/smj.v77i12.277048

All material is licensed under terms of the Creative Commons Attribution 4.0 International (CC-BY-NC-ND 4.0) license unless otherwise stated.

INTRODUCTION

Determining the identity of an unknown deceased is critical during a medicolegal investigation.1 A medical examiner and a forensic anthropologist work within the legal system to provide evidence and facilitate the identification of the deceased. This process ensures that justice is served to the individual and their family by resolving ambiguous questions, providing closure, and supplying accurate information for litigation purposes.2 In many instances, especially cold cases, the remains discovered are skeletonized. Hence, forensic anthropologists analyze the skeletal remains to establish a biological profile. The profile comprises sex, age, stature, and ancestry.3 Sex estimation is the initial step in the biological identification of the unknown as sex influences the other three parameters.4 Moreover, determining the sex minimizes the pool of candidates significantly by

50%.5

The human skeletal system is composed of several types of bones that can be categorized into long bones, flat bones, short bones, irregular bones and sesamoid bones. However, of all the bones in the human body, the bone that serve as the primary choice for anthropological sex estimation are a combination of irregular and flat bones; the pelvic (innominate) bone and the skull. The pelvic bone is primarily chosen for sex estimation as distinct morphological features can be witnessed between the sexes.6,7 Furthermore, the pelvis has been known to exhibit an accuracy of up to 95% in determining the

sex.8 Nonetheless, in the absence of the pelvic bone, the skull serves as the secondary choice in sex identification. This is because the accuracy has been observed to be approximately 92%.9 However, when unknown sets of remain are discovered, it is not always that a full set of skeletons is available. The incomplete nature of remains can be attributed to reasons such as animal scavenging, trauma, and the efficiency in retrieving and recovering the remains. Thus, in the context of forensic anthropology, different skeletal parts and bones are required to develop different methodologies in assessing the sex. One such bone is the patella.

The patella is the biggest sesamoid bone that is triangular in shape and possesses three borders – superior, medial and lateral, two surfaces – anterior and posterior, a base, and an apex.10 The patella is located within the patellofemoral groove of a femur and is embedded and attached to the quadricep femoris tendon and the patellar ligament.11 The posterior surface of the patella articulates with the femur forming an anterior articular surface of the knee joint that functions in protecting against trauma.2 Furthermore, the patella is also involved in the extensor mechanism of the quadricep muscle by acting as a fulcrum to increase the efficiency in the extension and flexion of the knee joint.12 Therefore, due to its morphology, location and function, the patella has been regarded as highly unsusceptible to post-mortem or taphonomy changes.13

In the past, patellae have been used in forensic cases to help identify unknown individuals. For example, in a 2008 case from Goias, Brazil,14 an unidentifiable decomposed body was discovered in a forest. Due to the severe state of decomposition, a fingerprint comparison was impossible. Alternatively, a positive identification was possible by comparing the patellae retrieved post-mortem to a radiographic image taken antemortem during knee surgery. The patella proved to be useful as the prior left knee surgery involving metallic surgical wires provided the evidence in the form of two channels in the patella that had exact dimensions to the surgical wires.14 Hence, this case showed that when a body is recovered, the conditions are not always favourable for identification. Several different parts of the skeleton should be used to determine the sex. In this instance, the patella has a location and position that allows it to resist post-mortem changes and is a useful tool in identifying the unknown. Thus, illustrates the importance of utilizing images to aid in distinguishing sex and identifying the individual Previously, numerous studies have assessed the sexual dimorphism of the patella. However, parameters regarded as the best predictors in sex estimation for both cranial and post cranial elements vary across populations, highlighting the importance of sexual dimorphism for each bone in specific population groups. While some studies have focused on morphometric measurments, others have incorporated machine learning for sex estimation.7,12 Nonetheless, many of these studies are limited by unequal15,16 or small sample sizes17,18, focus only on one side of the patella17 or restrict their method by primarily relying on discriminant function analysis for sex estimation.19 More recently, Cavlak et al20, have applied the framework of convolutional neural network to the MRI slices of patella, demonstrating the use of deep learning in sex estimation. As a result, building on these knowledge and foundation, this study aims to investigate sexual dimorphism using two different approaches; morphometric measurement inspected through logistic regression and machine learning, and photographic images evaluated through deep learning. In addition to addressing the comparative analysis, this studies aims to address limitation by using equal sample size and incorporating additional parameters. The contribution of this research is also in extending the image based analysis using photographs for a specific population. Therefore, it allows for a robust research by comparing traditional and advanced classification tools.

This is vital for forensic anthropologists to determine

the best method for estimating sex from the patella.

MATERIALS AND METHODS

Sample selection

This study analysed a total of 250 pairs of patellae in the Siriraj Anatomical and Anthropological Research Center’s (Si-AABRC) bone collection within the Department of Anatomy, Faculty of Medicine at Siriraj Hospital, Mahidol University. The sample consisted of an equal number of pairs of each sex, with an average age of 51 years. As both the left and right sides of the patella were used for the analysis, any remains without both patellae were excluded. Furthermore, patellae presented with fracture, erosion, trauma and extreme osteophytic formation were also excluded.

Parametric measurement

In this study, eight parameters in the patella were measured using a digital vernier calliper. The parameters were developed and adapted from the study conducted by Dayal & Bidmos19,21 assessing the sexual dimorphism of the patella in the South African population as well as the study assessing the sexual dimorphism of the patella in the Southern Part of Andhra Pradesh.17 The parameters shown in Fig 1. include maximum height (MAXH), maximum breadth (MAXB), maximum thickness (MAXT), height of the median ridge of the articular facet (HMRAF), breadth of lateral articular facet (BLAF), breadth of medial articular facet (BMAF), transverse diameter of articular facet (TDAF), and vertical diameter of the articular facet (VDAF). Detailed descriptions of the parameters are presented in Table 1. A patellar index (PI) was calculated to obtain a ratio between the MAXH and MAXL using the formula described below.

MAXH ×100

MAXB

Statistical analysis

The statistical analysis for the osteometric assessment of sexual dimorphism was conducted using Jamovi Version

2.5 - Sydney, Australia.22 The level of significance for each test was α = 0.05. Descriptive statistics comprising mean and standard deviation for all the parameters were calculated. The sample data were subjected to a Shapiro-Wilk normality test. If the data did not violate the assumption of normality, the difference between the left and right side of the patellae within each sex was assessed using independent t-test to determine the sexual dimorphism. However, if the data violated the aforementioned assumption, Welch’s t-test and Mann–Whitney U test were used, respectively. The percentage of sexual dimorphism was calculated using the formula

Fig 1. a) Patella measurements for the eight parameters. MAXH - Maximum Height, MAXB - Maximum Breadth, b) MAXT - Maximum Thickness, c) HMRAF - Height of median ridge of the articular facet, BLAF - Breadth of lateral articular facet, BMAF - Breadth of medial articular facet, d) VDAF - Vertical diameter of the articular facet, TDAF - Transverse diameter of the articular facet.

obtained from a study on sexual dimorphism of the root lengths in Tamil population.23

x

xmale - xfemale × 100

male

To assess the difference between the left and the right side of the patellae within each sex, the Wilcoxon Signed Rank (for data violating the normality) and paired t-test (for data following normal distribution) were used. Moreover, the mean difference was also calculated. With the presence of multiple parameters being analyzed, a Bonferroni and Benjamini–Hochberg test was also conducted and reported.

Reliability analysis by interobserver and intraobserver

In determining the reliability of the parameters, 20% of the samples (25 pairs of males and 25 pairs

of females) were randomly selected for interobserver and intraobserver evaluations. The intraobserver and interobserver measurements were collected two weeks apart. The Intraclass Correlation Coefficient (ICC) was calculated to discover the degree of correlation and reliability between two observers.24

Following the guidelines by Koo and Yi (2016)25 and Shrout and Fleiss (1979)26, intraobserver reliability was assessed using ICC(3,1), a two-way mixed-effects model and interobserver reliability was analyzed using (ICC(2,k), a two-way random-effects model. Both the upper and lower confidence intervals was reported. The ICC score were interpreted as follows: ICC < 0.50 is considered poor, 0.50 – 0.75 is moderate, 0.75 – 0.90 is good and >0.90 is considered excellent.

Logistic regression analysis

In the next step, a logistic regression equation was

TABLE 1. Description of the parameters measured on the patellae.

formulated using R Studio (Version 2024.4.2.764.1).27 The model was developed using the generalized linear model function – glm(). The output of this function represents a constant (a), a coefficient (b) and the measurement value of the predicting parameter (x) in the form of a + b × x.28 This model is then evaluated for the log of odd ratio (logit) and a probability of classification equation (P).

           1        

1+e-(a+b1.x1+b2.x2. )

The equation for the probability of classification predicts the outcome to be either male or female, where P > 0.5 classifies the measurement to be male and P ≤ 0.5 classifies the measurement to be female.29 A logit equation for each parameter and each side was developed. However, as there are multiple predictors, a stepwise model selection function was performed to obtain the parameter that provides the best efficacy for sex estimation based on an AIC, which is a computational value that depicts the efficacy of the model.28 The model that presented with the lowest AIC and highest Area Under the Curve (AUC) was selected. A ROC Curve as well as a Calibration plot was generated. For the model that provides the best efficacy,

a Hosmer–Lemeshow χ² alongside p-value and a Brier score was calculated. The model was then tested using a blind test where 100 additional pairs of the patellae were measured and the values were put in the model to determine the accuracy.

Machine learning

Machine Learning analyses were performed using MATLAB version 9.14 - 2023a30 and Waikato Environment for Knowledge Analysis software (WEKA 3.8.6).31 To prevent the subject-level leakage, each row in the dataset represented one subject containing the measurement of both left and right patella.

For analysis in WEKA, 250 subject pairs were randomly assigned to 10 groups, as the models were evaluated using 10-fold cross validation. During the training phase, the models were cross-validated using a separate test dataset selected through the ‘Supplied Test Set’ option. In each fold models were trained on 90% of the dataset and validated on the 10% of the dataset. The mean and standard deviation of the performance metrics were calculated across 10 folds.

For analysis in MATLAB, the response variable was set at “Sex”. The model training was then conducted

using the Classification Learner application with 10 fold cross validation. The mean and standard deviation of the performance metrics were reported for models exhibiting more than 80% of accuracy.

Deep learning

For the deep learning approach, 1000 images [left, right, anterior and posterior] were photographed using an iPhone 15 Pro at 3x magnification placed on an instrument (Fig 2) at a height of 24 cm. Each patella was secured with a clay to ensure a levelled orientation allowing the anterior surface and the posterior surface of the patella to be photographed (Fig 3). The deep learning assessment was done to only classify using anterior surface and only classify using posterior surface

Images in the dataset of the left and right patella

from the same subject were assigned exclusively to either training set (“Training → Male/Female”) or the validation test (“Test → Male/Female”) for both anterior images and posterior images, ensuring a subject-level split. Pretrained models like Visual Geometry Group (VGG19), MobileNet V2 and Deep Residual Learning (ResNet18) was utilized. The input images were resized to 224 pixels x 224 pixels with single precision and processed to 3 channel RGB. An augmentation was applied including random rotation (±10°), translation (±10 pixels), and scaling (0.9–1.1) in both axes.

Following this a 5-fold cross validation was utilized on the training dataset where each were set for 20 epochs, initial learning rate = 1 × 10-⁴ and batch size of 32. The sample was evaluated on a test set. Performance metrics was evaluated; accuracy, sensitivity and specificity.

a)

b)

Fig 2. a) A detailed schematic created using the Canva software showcasing the instrument used in taking images of the patellae. b) Images showcasing the set-up used for photographing the patella.

a) b)

Fig 3. a) Representative image of the anterior view of the patella. b) Representative image of the posterior view of the patella

RESULTS

Sexual dimorphism assessment and descriptive statistics Descriptive statistics consisting of mean and standard deviation (SD) were determined for all eight parameters for both sexes and sides. Comparing each parameter between males and females using parametric and non-parametric test, a p < 0.001 which is less than α = 0.05 was obtained. Seen in Table 2, the percentage of sexual dimorphism for each parameter is positive. The highest percentage and lowest percentage of sexual dimorphism were observed in TDAF (Left) at 12.15% and HMRAF (Right) at 6.72% respectively. Furthermore, the average PI on the left side of male was 95.97 and female was

95.82. The average patellar PI on the right side of both male and female were slightly smaller at 95.23 and 95.42 respectively. This is concurrent with the Bonferroni and Benjamini–Hochberg test where all parameters were significant. (q = 0.002) (Table 3).

Difference in patellae side assessment

Seen in Table 3, the results assessing the difference in the patellar side for each parameter for both males and females are summarized. All the parameters in females only the BMAF had a significant difference (p < 0.001), whereas in males MAXH (p < 0.001), TDAF (p = 0.041), and BLAF (p = 0.012) had a significant difference. However, from performing the Bonferroni and Benjamini–Hochberg only MAXH (q = 0.002) and BLAF (q = 0.020) showed significant difference within males and for females only BMAF (q = 0.002) (Table 3).

Reliability analysis by interobserver and intraobserver The ICC values for intraobserver and interobserver measurements on 50 randomly selected samples are summarized in Table 4. The intraobserver values ranged from 0.842 to 0.997 and the interobserver values ranged

from 0.502 to 0.995.

Regarding the both the reliability analysis, it was found that the highest ICC value belonged to the MAXB in both males and females. For the intraobserver value, the highest ICC was 0.995 for male on the left side and 0.997 for female on the right side. For the interobserver value, the highest ICC was 0.995 for male on the right side and 0.994 for both sides in female.

However, the parameter that proved to show the lowest reliability in interobserver measurement was the VDAF for both male and female as well as for both sides with values ranging from 0.502 to 0.706. The parameters that showcased the highest reliability for both male and female include MAXH, MAXB, and MAXT.

Logistic regression

The logistic regression analysis as seen in Table 5 indicates TDAF–transverse diameter of the articular facet on the left side as the most effective parameter in sex determination. This is because the model for TDAF produced the lowest AIC score of 212.20 and the highest AUC of 0.8941 (Table 5). Additionally, the ROC curve presented in Fig 5 visually demonstrates differences between the parameters, highlighting TDAF on the left side taking precedence in sex estimation.

TABLE 2. Descriptive statistics and percentage of sexual dimorphism of each parameter for both the sexes and both the sides for the patellae.

n = Sample size (pair of patellae)

a = Levene’s test (Homogeneity of variance test): violation of the assumption of equal variances. Welch’s t-test.

b = Shapiro Wilk (Normality Test): violation of the assumption of normality. Mann-Whitney U t-test.

Abbreviations: MAXH = Maximum Height, MAXB = Maximum Breadth, MAXT = Maximum Thickness, HMRAF = Height of the Median Ridge of the Articular Facet, BMAF = Breadth of Medial Articular Facet, BLAF = Breadth of Lateral Articular Facet, TDAF = Transverse Diameter of Articular Facet, VDAF = Vertical Diameter of Articular Facet.

TABLE 3. Comparison between the left and right sides for each parameter, separately for males and females.

a

L = Left and R = Right.

n = Sample size (pair of patellae)

a = Shapiro Wilk (Normality Test): violation of the assumption of normality. Wilcoxon signed-rank test

Abbreviations: MAXH = Maximum Height, MAXB = Maximum Breadth, MAXT = Maximum Thickness, HMRAF = Height of the Median Ridge of the Articular Facet, BMAF = Breadth of Medial Articular Facet, BLAF = Breadth of Lateral Articular Facet, TDAF = Transverse Diameter of Articular Facet, VDAF = Vertical Diameter of Articular Facet.

TABLE 4. Intraclass Correlation Coefficient (ICC) and descriptive analysis for the Intraobserver and Interobserver measurement.

Male Female

ICCa Lower CI Upper CI Mean SD ICCa Lower CI Upper CI

Mean SD

Parameters Side

MAXH L 42.5 4.31 0.981 0.962 0.990 37.1 2.17 0.994 0.989 0.997

MAXB

L

R

1. 3.92 0.995 0.990 0.998 39.0 2.89 0.996 0.993

2. 3.97 0.993 0.987 0.997 39.3 2.73 0.997 0.994

0.998

0.998

R 42.1 4.41 0.984 0.969 0.992 37.3 2.13 0.993 0.986 0.996

MAXT L 20.3 1.93 0.986 0.973 0.993 18.5 1.30 0.982 0.965 0.991

HMRAF

L

R

26.6 3.22 0.916 0.840 0.957 24.8 2.24 0.873 0.766

27.2 2.94 0.951 0.906 0.975 24.7 2.42 0.903 0.819

0.933

0.949

R 20.1 1.77 0.992 0.984 0.996 18.6 1.28 0.957 0.917 0.978

BMAF L 20.8 1.77 0.861 0.741 0.927 18.3 1.78 0.894 0.802 0.944

BLAF

L

R

26.5 2.26 0.927 0.859 0.962 23.9 1.77 0.863 0.749

26.4 2.78 0.960 0.922 0.979 23.8 1.79 0.842 0.713

0.928

0.916

R 20.9 1.83 0.906 0.821 0.951 19.1 1.58 0.852 0.729 0.921

TDAF L 42.7 3.74 0.982 0.965 0.991 37.6 2.62 0.973 0.948 0.986

R 42.5 3.69 0.980 0.961 0.990 37.7 2.36 0.938 0.881 0.968

MAXB L 43.5 3.76 0.992 0.977 0.996 38.8 2.85 0.994 0.981 0.997

MAXT

L

R

19.9 1.91 0.973 0.830 0.991 18.3 1.31 0.980 0.895

19.8 1.81 0.970 0.890 0.988 18.4 1.38 0.956 0.786

0.992

0.983

R 43.7 3.90 0.995 0.988 0.998 39.1 2.60 0.994 0.973 0.998

HMRAF L 25.3 3.46 0.867 0.462 0.949 23.6 2.05 0.781 0.128 0.917

BMAF

L

R

19.6 1.91 0.874 0.499 0.951 17.4 1.75 0.898 0.406

19.5 1.71 0.791 0.158 0.921 18.2 1.57 0.820 0.383

0.964

0.927

R 25.4 3.26 0.876 0.036 0.963 23.6 2.29 0.786 0.220 0.915

BLAF L 25.9 2.54 0.911 0.588 0.967 23.6 1.92 0.889 0.472 0.959

TDAF

L

R

41.4 3.70 0.951 0.571 0.984 36.5 2.57 0.910 0.110

41.0 3.49 0.923 0.262 0.976 37.1 2.69 0.931 0.411

0.973

0.978

R 25.8 2.73 0.949 0.563 0.984 23.6 1.91 0.911 0.819 0.955

VDAF L 28.5 2.35 0.706 -0.112 0.898 26.6 1.90 0.502 -0.142 0.791

R 28.8 2.23 0.665 -0.125 0.879 26.6 2.0 0.586 -0.140 0.839

a = Intraclass Correlation Coefficient for the comparison between the intraobserver measurement and original as well as, interobserver measurement and original

Abbreviations: MAXH = Maximum Height, MAXB = Maximum Breadth, MAXT = Maximum Thickness, HMRAF = Height of the Median

Ridge of the Articular Facet, BMAF = Breadth of Medial Articular Facet, BLAF = Breadth of Lateral Articular Facet, TDAF = Transverse Diameter of Articular Facet, VDAF = Vertical Diameter of Articular Facet.

TABLE 5. Logistics Regression Model for single and combination of parameters for sex estimation.

AUC

Odds Ratio (95% CI)

AIC

Logit

Side

Parameters

MAXH L -19.993 + 0.505 ×MAXHL 230.67 1.66 (1.48 – 1.89) 0.8652

MAXB

L

R

-21.415 + 0.516×MAXBL

-19.701 + 0.475×MAXBR

227.33

237.55

1.68 (1.49 – 1.92)

1.61 (1.44 – 1.83)

0.8786

0.8634

R -20.120 + 0.512×MAXHR 236.92 1.67 (1.48 – 1.91) 0.8607

MAXT L -15.894 + 0.809×MAXTL 263.66 2.25 (1.83 – 2.83) 0.8210

HMRAF

L

R

-7.190 + 0.280×HMRAFL

-7.030 + 0.272×HMRAFR

319.27

321.28

1.32 (1.19 – 1.48)

1.31 (1.18 – 1.47)

0.6977

0.6921

R -15.497 + 0.791×MAXTR 267.27 2.21 (1.80 – 2.77) 0.8190

BMAF L -11.445 + 0.594×BMAFL 276.02 1.81 (1.55 – 2.16) 0.8062

BLAF

L

R

-20.209 + 0.795×BLAFL

-17.549 + 0.697×BLAFR

232.64

250.69

2.22 (1.84 – 2.74)

2.01 (1.70 – 2.43)

0.8709

0.8458

R -11.089 + 0.566×BMAFR 286.98 1.76 (1.51 – 2.09) 0.7835

TDAF L -22.917 + 0.571×TDAFL 212.20 1.77 (1.56 – 2.05) 0.8941

R -20.223 + 0.505×TDAFR 229.50 1.66 (1.48 – 1.89) 0.8767

0.76 0.84 0.79 0.81 0.80

Fig 4. A confusion matrix representing the performance evaluation of the best model obtained through logistic regression analysis.

Fig 5. Receiver operating characteristic curve (ROC curves) for the all the univariable measured on the patella.

Moreover through stepwise logistic regression, the most notable model for sex estimation incorporated the parameters MAXH(R), BLAF(L), TDAF(L), formulating the equation: -26.87+(TDAF.L×0.277)+(MAXH.

R×0.219)+(BLAF.L×0.281) (Table 5). However, within these three the parameter that contributes most to the prediction is MAXH(R) and TDAF(L) as their p values are statistically significant at 0.007 and 0.012 respectively. The analysis resulted in an AIC score of 205 and an AUC of 0.907 (95% CI: 0.8685 - 0.945) (Fig 6). The performance metrics revealed a sensitivity of 0.864, specificity of 0.808,

a precision of 0.818, NPV of 0.856 and accuracy of 0.836 (95% CI: 0.7842 - 0.8797) (Table 5). A Hosmer–Lemeshow test revealed X2 of 8.2406 with a p-value of 0.4103 – an appropriate goodness of fit. In addition, the Brier score revealed a low prediction error with a value of 0.12.

Machine learning and deep learning

Different models were obtained from using WEKA software and MATLAB software that exhibited the accuracy of 80% or more (Table 6). The models in MATLAB that produced the highest were Coarse Tree and Coarse

a)

b)

c)

Fig 6. a) Receiver operating characteristic curve (ROC curves) for the equation that best determines sex in patella b) Receiver operating characteristic curve (ROC curves) with confidence band. c) Calibration plot for the equation that best determines sex in patella: TDAF.L + MAXH.R + BLAF.L.

TABLE 6. Comparing the top ten different machine learning models from WEKA and MATLAB to determine the model with the highest accuracy in sex estimation.

TABLE 7. Deep Learning Models classifying male and female patellae.

Gaussian SVM at 83.7% while the lowest was Weighted KNN at 80.4%. On the other hand, in WEKA the model with the highest validation accuracy is Jrip and OneR with 86.0% and the lowest accuracy is Logistic (81.2%). Hence, both WEKA and MATLAB provide models with similar ranging validation accuracy.

In addition, from the three models that were trained and tested in classification of images generated the best results with the accuracy of 79% for the ResNet18. Whereas the lowest accuracy obtained was for VGG19 with an accuracy of 49.7%.

DISCUSSION

The aim of this investigation was to conduct a comparative analysis of several sex estimation modalites and evaluate the efficacy of deep learning model to machine learning and logistic regression for sex estimation using the dry patellae. By combining morphometric and image-based analysis, this paper extends traditional statistical models and discriminant functions. The results obtained help reinforce that patella showcases apparent sexual dimorphism and can provide accuracy in the absence of the primary skeleton choice (pelvis and skull).

Furthermore, the advantage of using patella can be attributed to both the external factors and internal factors like hormones, genes and biomechanics that influence the osseous robustivity within each sex.32 Therefore, other bones such as the patella can be of an alternative. The anatomical position, dense33 composition and the articular cartilage of the patellae is composed of copious amounts of collagen and glycosaminoglycan34, making it resilient to taphonomic changes. As a result it increases the probability of recovery during excavation.

Sexual dimorphism assessment and descriptive statistics In this study, the morphometric examination confirmed that there is a significant difference between all eight parameters between males and females, which is consistent with several previous studies on the sexual dimorphism of this bone. Furthermore, the dimensions of

the patella were larger in males than female, attributing to the hormonal factor like oestrogen which has a stimulatory effect to the increase in the periosteal growth in males but contrary in females, acts as inhibitory effect on the periosteal growth.35 These results are in accordance with the findings of several previous researches conducted on different populations such as South African Whites19, South African of European ancestry18, Central Thai population10, Northern Thai population15, Spanish population13, Southern Part of Andhra Pradesh, India17 and Southern Italian population.36

Comparing the average between sexes within regional boundaries, the patella in this study exhibited a larger value for the HMRAF and BMAF in comparison to the Northern Thai region15 and Central Thai region.10 The hypothesis for this could be variation within a biology of a subgroup of a population but also it could be the indefinite landmark on the posterior aspect of the patella. Furthermore, comparing the mean values of this study to the findings from Southern Chinese population37 and Korean population (total knee arthroplasty)38 can infer that Asian population may have smaller patellar dimensions compared to Westerners, although there has also been a study on the Asian population with healthy knees that revealed no difference between Asians and Westerners.39 The discrepancy in the measurement can be attributed to to the different genetic makeup, lifestyle and environment.15

In the forensic context, the differences in these

datahighlight the importance of using population specific reference. Furthermore, with the presence of apparent sexual dimorphism, the patella could be used for sex estimation in the field of forensic anthropology. In the area where remains are found, if the primary bones are not available, patella can be the supplementary element to help determine the sex.

Difference in patellae side assessment

Furthermore, most of the findings in our study were consistent to both sides of the patella, indicating that any

side can be used in sex estimation. However, minor side specific difference was observed in BMAF for females and MAXH and BLAF in males. While these finding contradicts the information from previous research on Portuguese population16 and Spanish population13 reporting no significant difference, one plausible reason could be the existence of bilateral variation. Nevertheless, the differences in skeletal measurements may be limited or inconsistent across different parameters, suggesting a minimal impact. In addition, the variation of these significant differences between left and right side lacks comparative studies on the bilateral factor of patella, suggesting a need for further research.

Reliability analysis by interobserver and intraobserver The ICC values for the interobserver and intraobserver measurement ranged from 0.502 to 0.995 and 0.842 to 0.997 respectively. The parameters that showcased the highest reliability for both male and female include MAXH, MAXB, and MAXT because these parameters have definite landmarks with methodology being repeated by many authors. In addition, research on the Thai population by Srinak and Sukvitchai10 reported similar results. This reinforces the repeatability of the methodology being used. However, the lowest interobserver measurement can be attributed to the idea that even though the observers were using the same description, the inconsistency and subjectivity of the landmarks can cause the discrepancies. The VDAF parameter had the lowest value and this could be due to the posterior side of the patella having a smooth surface. Therefore, the margins are not as defined compared to the other parameters. Moreover, this also further suggest that parameters such as VDAF should probably be omitted when estimation sex of an

unidentified individual.

Logistic regression

In the field of forensic science, two of the most common methods that are used in formulating a sex estimation equation are logistic regression and discriminant function analysis (DFA).41 Thus, unlike most of the previous studies on sexual dimorphism on the patella that uses DFA, this study performed a logistic regression analysis to evaluate the parameters that estimate sex. The reason for this is that the data obtained in this study have parameters with measurements that are not normally distributed. Hence, logistic regression is more suitable because it can provide robust estimations even when assumptions are violated.41 Additionally, logistic regression offers greater flexibility compared to DFA because the generated equation can be directly utilized

to compute the probabilities of categorizing the sexes, thereby quantifying uncertainties. In contrast, DFA solely provides the equation and cut-off point, resulting in the absence of prior probabilities. This absence diminishes the confidence in classifying the sexes, potentially leading to misclassification.42

Many previous studies have reported that the measurement that exhibited the highest sexual dimorphism varied among populations as it included MAXH (South African White, Portuguese, Spanish)13,16,19 MAXB (Portuguese)16, MAXT (Italian and Spanish)13,36, BLAF (Central Thai and Northern Thai).10,15 However, in this study, TDAF was most sexually dimorphic with 12.15% on the left side and 11.53% on the right side. This is further iterated by the logistic regression analysis, producing the highest AUC but the lowest AIC value. TDAF is not a common parameter measured in previous studies but rather divided into the breadth of medial facet (BMAF) and breadth of lateral facet (BLAF). Furthermore, a stepwise regression analysis revealed the parameters identified as the most notable for sex estimation were MAXH.R, BLAF.L, and TDAF.L. The performance metrics for this LR model revealed a strong predictability model. With high sensitivity and specificity value as well as precision and negative predictive value, the model is very effective in accurately distinguishing between the sexes. A blind test of 100 samples revealed an accuracy 80.0%. Thus, these results are comparable with other populations such as the 73% - 89%.10,13,15,19,21 The value obtained in this study is in the similar range of previous studies, but the methodologies used to obtain them were different, as all of these studies generated discriminant function equations. Nonetheless, as each population has different genetic composition, physical activities and biomechanics, it is important to obtain equations for each population. However, this study also possess some limitations like the lack of external validation. Having another collection for testing would help reinforce the accuracy. This would be a possible next step for future projects.

Studies conducted for sex estimation using the most

ideal equation in each population were applied to different populations by different authors13,19, revealed a lower accuracy in classifying the sexes.16 Therefore, from the LR analysis, it can be suggested that patella provides an important significance in determining sex. It also signifies the importance of population specific determinator. This also provides the forensic anthropologist an alternative approach when profiling an unknown remain.

While, the logistic regression yielded a strong predictive value of 80%, the performance is lower compared

to pelvis and skull at approximately 92% to 95%. Thus, this suggest that patella while it can fulfill the role of being a supplementary bone for sex estimation, it cannot replace the primary models. Locally and internationally, the findings from studies conducted on patella has expressed it to be a viable bone in aiding the sex estimation procedure but only when the primary alternatives are missing, fragmented or unsuccessfully preserved.

Machine learning and deep learning

Logistic regression and discriminant function analysis are the prevailing methods in sex estimation but as times have been changing and there is an integration of artificial intelligence into every aspect of our daily lives. Hence, there has also been a growing tendency to use artificial learning models such as ML and DL in the field of forensic science.12

Previously, there have been two studies that utilized a ML approach predicting the best model for sex estimation: Knecht and colleagues12 for an Italian population and Bidmos and colleagues18 for a South African population with mixed European and African descent.

Comparing this current research with the work of Bidmos and colleagues7, they achieved an overall accuracy of 89.61% using Random Forest, 89.23% using AdaBoost classifier and 90.77% using the Stacking model. The approach to the training model was different as a 5-fold cross validation training was used whereas this study used 10-fold cross validation. Using 5-fold cross validation reduced the overall accuracy for each classifier. Comparing this with the Italian population12, their overall accuracy for their ML classifiers ranged from 85% to 91% for training data and 71% to 95% for their validation sample. The best classifier for their study was the Sequential minimal optimization (SMO) which is a type of support vector machine (SVM). Furthermore, their research only utilizes three parameters in classifying sex whereas, this one uses 8 parameters, broadening the horizon of computing differences. Nonetheless, the overall accuracies are in similar ranges showcasing the agreement of using ML as a means for classifying sex. Not only this, it further strengthens the applicability of using patella in sex estimation.

Furthermore, depending on the population, various

classifiers would provide a different overall accuracy. This suggests that to study the Thai population, an ML model such as Coarse tree or SVM can be used.

Extending a step further, utilizing the images of the patella for sex estimation provides a flexible alternative to the primary choice of pelvis or the skull. ResNet 18 likely gained the highest accuracy because it is composed

of residual network that help in training at deeper layer and this allows for information to skip layers allowing the model to learn more reliably.43 However, VGG19 obtained the lowest accuracy because this is model is dense with large parameters and layers that could highly lead to overfitting on small dataset like 1,000 images.44 Finally, the MobileNetV2 is a lightweight model which could reduce flexibility in trying to read complex images.45 However, looking at the results from a different perspective, the lower accuracy value can be ascribed to the type of image used in training and testing. The anterior and posterior images used in this study as seen in Fig 3 can be described to have a cluttered or noisy background. Prior studies have reported that training and testing a convolutional network with images with cluttered background can slower the training procedure and cause models to underperform.43 Thus, this can act as a limitation not only for this study but a cautious point for future researches. The lower accuracy for our study can possibly be explained by a small dataset of 1000 images because in recent research using convolutional neural network to estimate sex from MRI images of patella, they utilized approximately 6,000 images.20 Hence, a small dataset increases the risk of overfitting and weakens the generalizability factor. An MRI image is also more homogenous with a reduced background cluttering.

Thus, these differences could also explain discrepancies

and lower result yeilded.

The selection of models like the VGG19, ResNet18 and MobileNetV2 was to study a range of model. Not specifically for forensic anthropology but these models have been used to study age and sex estimation from dental radiograph in young adults, achieving an accuracy of approximately 80%.44

Computational forensics is a current developmental content that utilizes AI to identify human remains. Deep learning analysis usually employs radiographic datasets. Studies that have utilized CT scans37 and MRI45 in measuring the patella size to determine sexual dimorphism have been able to achieve high accuracy. Furthermore, another relevant study on using DL to classify sex in relation to the patella is the radiographs of tibiofemoral joints, that obtained 90.3%.46 However, this study demonstrated a more flexible approach by using photographic imagery which is an easier, practical and much more cost-effective manner. However, with the limited sample size of 1,000 images the training and model can be developed to achieve higher accuracy. It is very important to always note the sample size, population and the methodologies utilized in each study before performing any comparison.

CONCLUSION

In conclusion, this study of sexual dimorphism of the patella within the Thai population using 250 paired patellae revealed sexual dimorphism exists in the patella as all measured parameters showed significant differences. Overall, comparing the traditional approach of logistic regression to advanced computational models such as ML and DL, it is evident that all these methods achieve an accuracy of 80% and above. These findings suggest that while logistic regression can be the basis of generating a robust sex estimation equation, ML and DL models can work to enhance the classification process. This suggests that even in the absence of the primary skeletal choice for sex determination (pelvis and skull), the patella can be used.

Data Availability Statement

The access of this data was approved through Siriraj Anatomical and Anthropological Bone Research Center (Si-AABRC) procedure. Data is available on reasonable request from Si-AABRC.

ACKNOWLEDGEMENTS

The authors of this research are thankful for Siriraj Anatomical and Anthropological Bone Research Center (Si-AABRC), Department of Anatomy, Faculty of Medicine Siriraj Hospital for allowing the authors to utilize the patellae from the bone collection. The authors are also grateful for the body donation program of Department of Anatomy, Faculty of Medicine Siriraj Hospital. This research would like to acknowledge Ms. Areeya Marie

B. Wongla for her help.

DECLARATIONS

Grants and Funding Information

None.

Conflict of Interest

None.

Registration Number of Clinical Trial

Not Applicable.

Author Contributions

Conceptualization and methodology. S.H.; Investigation, S.H.; Formal analysis, S.H. P.R., P.J.,N.C.; Visualization and writing – original draft, S.H.; Writing – review and editing, P.R., N.C., P.J., S.R., R.M., N.S.; Supervision, N.S. All authors have read and agreed to the final version of the manuscript.

Use of Artificial Intelligence

This paper utilized ChatGPT-5 (Open AI) to generate initial code drafts for R-Studio and MATLAB.

ETHICAL STATEMENT

This research was approved by the Siriraj Institutional Review Board (SiRB), Faculty of Medicine Siriraj Hospital [SIRB Protocol No.331/2568 (IRB4)].

REFERENCES

1. Kahana T, Hiss J. Identification of human remains: forensic radiology. J Clin Forensic Med. 1997;4(1):7–15.

2. de Boer HH, Blau S, Delabarde T, Hackman L. The role of forensic anthropology in disaster victim identification (DVI): recent developments and future prospects. Forensic Sci Res. 2018;4(4): 303-15.

3. Austin D, King RE. The Biological Profile of Unidentified Human Remains in a Forensic Context. Acad Forensic Pathol. 2016;6(3):370–90.

4. Krishan K, Chatterjee PM, Kanchan T, Kaur S, Baryah N, Singh RK. A review of sex estimation techniques during examination of skeletal remains in forensic anthropology casework. Forensic Sci Int. 2016;261:165.e1–8.

5. Ahmed AA. Estimation of sex from the lower limb measurements of Sudanese adults. Forensic Sci Int. 2013;229(1-3):169.e1–7.

6. Phenice TW. A Newly Developed Visual Method of Sexing the Os Pubis. Am J Phys Anthropol. 1969;30(2):297–301.

7. Bidmos MA, Olateju OI, Latiff S, Rahman T, Chowdhury MEH. Machine learning and discriminant function analysis in the formulation of generic models for sex prediction using patella measurements. Int J Legal Med. 2023;137(2):471–85.

8. Memarian A, Aghakhani K, Mehrpisheh S, Fares F. Gender determination from diagnostic factors on anteroposterior pelvic radiographs. J Chin Med Assoc. 2017;80(3):161–8.

9. Wangdee A, Thipdet W, Prasitwattanaseree S, Singsuwan P, Mahakkanukrauh P. Efficiency of sex determination by using external morphology of the pelvis in Thai population. BSCM. 2014;53(4):175–9.

10. Srinak N, Sukvitchai P. Sex estimation from patella using discriminant analysis in Central Thai population. Journal/ Journal - Canadian Society of Forensic Science. 2023;56(4):231–47.

11. Fox A, Wanivenhaus F, Rodeo S. The Basic Science of the Patella: Structure, Composition, and Function. J Knee Surg. 2012;25(2): 127–41.

12. Knecht S, Morandini P, Biehler-Gomez L, Nogueira L, Adalian P, Cattaneo C. Sex estimation from patellar measurements in a contemporary Italian population: a machine learning approach. Int J Legal Med. 2024;139(3):1371–80.

13. Peckmann TR, Meek S, Dilkie N, Rozendaal A. Determination of sex from the patella in a contemporary Spanish population. J Forensic Leg Med. 2016;44:84–91.

14. Silva RF, Franco A, Santos D, Picoli FF, Marinho DE de A. Human identification through the patella—Report of two cases. Forensic Sci Int. 2014;238:e11–4.

15. Chompoophuen H, Tipmala J, Duangto P, Mahakknukrauh

    P. Sex Determination from the Patella in a Thai Population. Int J Morphol. 2024;42(4):891–7.

    
    

16. Maio C, Cunha E, Navega D. Metric analysis of the patella for sex estimation in a Portuguese sample. Forensic Sci Res. 2024; 9(2):owae015.

17. Nagarjuna K, Mamatha K, Venkateswarlu B. Patellar Anthropometry in Sex Differentiation - A Study in the Southern Part of Andhra Pradesh, India. Indian Journal of Forensic Medicine and Toxicology. 2021;15(2):3113–8.

18. Olateju OI, Philander I, Bidmos MA. Morphometric analysis of the patella and patellar ligament of South Africans of European ancestry. South African Journal of Science. 2013;109(9/10):1–6.

19. Bidmos MA, Steinberg N, Kuykendall KL. Patella measurements of South African whites as sex assessors. Homo. 2005;56(1):69–74.

20. Cavlak N, Çınarer G, Erkoç MF, Kılıç K. Sex estimation with convolutional neural networks using the patella magnetic resonance image slices. Forensic Sci Med Pathol. 2025;21(2):628–39.

21. Dayal MR, Bidmos MA. Discriminating Sex in South African Blacks Using Patella Dimensions. J Forensic Sci. 2005;50(6): 1294-7.

22. The jamovi project (2024). jamovi (Version 2.5) [Computer Software]. Available from: https://www.jamovi.org

23. Govindaram D, Bharanidharan R, Ramya R, Rameshkumar A, Priyadharsini N, Rajkumar K. Root Length: As a determinant tool of sexual dimorphism in an ethnic Tamil population. J Forensic Dent Sci. 2018;10(2):96-100.

24. Bobak CA, Barr PJ, O’Malley AJ. Estimation of an inter-rater intra-class correlation coefficient that overcomes common assumption violations in the assessment of health measurement scales. BMC Med Res Methodol. 2018;18(1):93.

25. Koo TK, Li MY. A Guideline of Selecting and Reporting Intraclass Correlation Coefficients for Reliability Research. J Chiropr Med. 2016;15(2):155–63.

26. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86(2):420–8.

27. RStudio Team (2020). RStudio: Integrated Development for

    R. RStudio, PBC, Boston, MA, USA. Available from: http:// www.rstudio.com

28. Verma R, Krishan K, Rani D, Kumar A, Sharma V, Shrestha R, et al. Estimation of sex in forensic examinations using logistic regression and likelihood ratios. Forensic Science International: Reports. 2020;2:100118.

29. Dietrichkeit Pereira JG, Fróes Lima K, Alves da Silva RH. Mandibular Measurements for Sex and Age Estimation in Brazilian Sampling. Acta Stomatologica Croatica. 2020;54(3):294–301.

30. The MathWorks Inc. MATLAB version: 9.14 (R2023a), Natick, Massachusetts: The MathWorks Inc., 2023. Available from: https://www.mathworks.com

31. Frank E, Hall MA, Witten IH. The WEKA Workbench. Online Appendix for "Data Mining: Practical Machine Learning Tools and Techniques", Morgan Kaufmann, Fourth Edition, 2016.

32. Sangchay N, Dzetkuličová V, Zuppello M, Chetsawang J. Consideration of Accuracy and Observational Error Analysis in Pelvic Sex Assessment: A Study in a Thai Cadaveric Human Population. Siriraj Med J. 2022;74(5):330–9.

33. Abdelaziz S, Khattab HM, AbdelHaq NA. Prediction of Sex

    from Patellar Parameters Obtained by Magnetic Resonance Imaging in a Sample of Adult Egyptians. Zagazig Journal of Forensic Medicine and Toxicology. 2024;22(2):1–16.

34. Kim Y-M, Joo Y-B. Patellofemoral Osteoarthritis. Knee Surg Relat Res. 2012;24(4):193–200.

35. Nieves JW, Formica C, Ruffing J, Zion M, Garrett P, Lindsay R, et al. Males Have Larger Skeletal Size and Bone Mass Than Females, Despite Comparable Body Size. J Bone Miner Res. 2004;20(3): 529–35.

36. Introna F, Di Vella G, Campobasso CP. Sex determination by discriminant analysis of patella measurements. Forensic Sci Int. 1998;95(1):39–45.

37. Shang P, Zhang L, Hou Z, Bai X, Ye X, Xu Z, et al. Morphometric measurement of the patella on 3D model reconstructed from CT scan images for the southern Chinese population. Chin Med J (Engl). 2014;127(1):96–101.

38. Kim MH, Yoo MJ, Seo JB, Park HG, Shim SH. Statistical Analysis of the Patellar Thickness in Adults by MRI. J Korean Orthop Assoc. 2005;40(6):646-51.

39. Pontoh LA, Dilogo IH, Rahyussalim AJ, Widodo W, Edwin R, Fiolin J. Evaluation of patellar dimension and Bristol Index in Asian population: AnMRIstudy. AnnMedSurg(Lond). 2021;72:103072–2.

40. Kanchan T, Menezes RG, Moudgil R, Kaur R, Kotian MS, Garg RK. Stature estimation from foot dimensions. Forensic Sci Int. 2008;179(2-3):241.e1–5.

41. Bartholdy BP, Sandoval E, Hoogland MLP, Schrader SA. Getting Rid of Dichotomous Sex Estimations: Why Logistic Regression Should be Preferred Over Discriminant Function Analysis. J Forensic Sci. 2020;65(5):1685–91.

42. Antonogeorgos G, Panagiotakos DB, Priftis KN, Tzonou A. Logistic Regression and Linear Discriminant Analyses in Evaluating Factors Associated with Asthma Prevalence among 10- to 12-Years-Old Children: Divergence and Similarity of the Two Statistical Methods. Int J Pediatr. 2009;2009:952042.

43. Kc K, Yin Z, Li D, Wu Z. Impacts of Background Removal on Convolutional Neural Networks for Plant Disease Classification In-Situ. Agriculture. 2021;11(9):827–7.

44. Adedigba AP, Adeshina SA, Aina OE, Aibinu AM. Optimal hyperparameter selection of deep learning models for COVID-19 chest X-ray classification. Intell Based Med. 2021;5:100034.

45. Kalaiselvi K, Kasthuri M. Tuning VGG19 hyperparameters for improved pneumonia classification. The Scientific Temper. 2024;15(02):2231–7.

46. Ukwandu O, Hindy H, Ukwandu E. An evaluation of lightweight deep learning techniques in medical imaging for high precision COVID-19 diagnostics. Healthc Anal (N Y). 2022;2:100096.

47. Kim HE, Cosa-Linan A, Santhanam N, Jannesari M, Maros ME, Ganslandt T. Transferlearning formedical image classification: a literature review. BMC Med Imaging. 2022;22(1):69.

48. Yoo JH, Yi SR, Kim JH. The geometry of patella and patellar tendonmeasuredon Knee MRI. Surg Radiol Anat. 2007;29(8):623–8.

49. Oura P, Junno JA, Hunt DR, Lehenkari P, Tuukkanen J, Maijanen H. Deep learning insex estimation from knee radiographs – A proof-of-concept study utilizing the Terry Anatomical Collection. Leg Med (Tokyo). 2023;61:102211.

‌In Silico Study of Functional Network Analysis Focusing on Antipsychotic Drugs and Cognitive Function in Schizophrenia

Kamonwan Thanontip, M.Sc.1, Banthit Chetsawang, Ph.D.1, Vorasith Siripornpanich, M.D., Ph.D.1, Vasunun Chumchua, D.Sc.2, Jirapa Chetsawang, M.D., Ph.D.3,*

1Research Center for Neuroscience, Institute of Molecular Biosciences, Mahidol University, Salaya, Nakhon Pathom, Thailand, 2National Institute

for Child and Family Development, Mahidol University, Salaya, Nakhon Pathom, Thailand, 3Department of Anatomy, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand.

*Corresponding author: Jirapa Chetsawang E-mail: jirapa.cht@mahidol.ac.th

Received 16 August 2025 Revised 25 October 2025 Accepted 25 October 2025 ORCID ID:http://orcid.org/0000-0002-7776-6456 https://doi.org/10.33192/smj.v77i12.277103

All material is licensed under terms of the Creative Commons Attribution 4.0 International (CC-BY-NC-ND 4.0) license unless otherwise stated.

ABSTRACT

Objective: Recent evidence has emphasized cognitive dysfunction in psychiatric disorders. This study explores bioinformatics analysis aimed at investigating gene targets and pathways related to antipsychotic drug effects on cognitive function in schizophrenia. The main purpose of conventional antipsychotics is to treat the positive symptoms, but they have limited evidence for the involvement of these drugs in cognitive function. Therefore, this study aimed to explore the pathway for the antipsychotic signaling in the cognitive function of schizophrenia. Materials and Methods: Nine frequently used antipsychotics were included from the literature review. The GeneCards database was utilized to identify the drug target genes associated with cognitive function in schizophrenia. The gene lists were analyzed with two different tools, WebGestalt and STRING. The overlapping genes that result from the two previous steps may play an essential role in cognitive function. These genes were analyzed to find out the related pathway using the KEGG database.

Results: The potential gene list consists of CHRM1, HRH3, DRD2, GRM5, HTR2A, and SLC6A1. The functional

enrichment in KEGG reveals six target proteins involved in several pathways, which play an essential role in cognitive function, including the neuroactive ligand-receptor interaction, the gap junction, the calcium signaling pathway, and the cAMP signaling pathway.

Conclusion: The results reveal that G-protein-coupled receptors (GPCRs) are the main target site of potential

mechanisms or processes related to antipsychotics and cognitive function. Therefore, GPCRs might be a promising candidate for future research on potential therapeutic targets for intervention in neurological dysfunction-associated cognitive deficits in schizophrenia.

Keywords: In silico study; functional network analysis; antipsychotic drugs; cognitive function; schizophrenia (Siriraj Med J 2025; 77: 847-857)

INTRODUCTION

Schizophrenia is a psychiatric illness characterized by psychotic symptoms, including positive symptoms and negative symptoms.1,2 Cognitive deficit is a neurological dysfunction that occurs in schizophrenia.3-6 The cognitive domains that show differential impairment in schizophrenia include attention, executive function, language skills, working memory, and cognitive flexibility.7,8 These determine the functional impairment characteristic of schizophrenia, which disturbs their daily life activity and quality of life.9 Therefore, the most concerning clinical target in the treatment of schizophrenia is not only to reduce the psychotic symptoms but also to improve cognitive function.10

Cognitive impairment may develop from the pathology of the disease itself in some schizophrenia patients.11 It has been implicated that conventional antipsychotics are effective in treating psychological symptoms; however, they cannot improve cognitive function.12 The effects of antipsychotics on cognition are complex, which might depend on the stage of illness and the doses of antipsychotics. It has been reported that a dose reduction in a patient who is clinically in an advanced phase of remission, causing cognitive decline associated with antipsychotic drugs. On the other hand, dose reduction or discontinuation after reaching stable remission might

improve cognitive function.13 Several lines of evidence revealed that the usage of antipsychotics in high doses impairs cognitive function in schizophrenia patients.14,15 On the contrary, a low dose of conventional medications such as haloperidol improves cognitive functions.16

According to the heterogeneity of antipsychotics in their pharmacologic properties, they have differential profiles of cognitive efficacy in patients with schizophrenia. Therefore, understanding the molecular mechanisms and mode of action of each antipsychotic related to cognitive processes is needed to address the gap in knowledge. Computational network analysis is a potential tool that may answer these knowledge gaps by providing a biological explanation for antipsychotic targets and insights for the putative molecular mechanism of antipsychotic agents, focusing on cognitive function. Therefore, this study aimed to explore the interaction of the target protein, which is related to the action of antipsychotic drugs and cognitive functions related to schizophrenia. The results of this study could be explored to discover more effective novel target treatment of neurological dysfunction involving neuropsychological-associated cognitive deficit.

MATERIALS AND METHODS

The conceptual framework was designed to gain

insight into the pathway of the potential protein involved in antipsychotics and cognitive function in schizophrenia. The framework starts to identify the commonly used antipsychotics for schizophrenia using a literature review and then using bioinformatics tools to identify gene targets for selected antipsychotics and genes associated with cognitive function in schizophrenia. Next step, the gene associated with antipsychotics and cognitive function in schizophrenia was identified. Then, protein-protein interaction network analysis and network clustering were performed, followed by the identification of a common gene that is associated with antipsychotics and cognitive function in schizophrenia from two different methods. Finally, the pathway that contains potential protein involves antipsychotics and cognitive function in schizophrenia was identified (Fig 1).

Identification of commonly used antipsychotics

In silico study of network analysis focusing on antipsychotic drugs and cognitive function in people with schizophrenia was started to identify commonly used antipsychotics for schizophrenia (step 1 of Fig 1) by reviewing published articles and textbooks.17-20

Fig 1. Conceptual framework of experimental design.

Identification of gene targets for selected antipsychotics

GeneCards (https://genecards.org) and a database for searching genes of interest (step 2 of Fig 1) were used to identify gene targets for selected antipsychotics. Target genes of selected antipsychotics were identified using the names of 9 antipsychotics as a keyword. The keyword for searching in this step is “[compounds] (Aripiprazole) OR [compounds] (Chlorpromazine) OR [compounds] (Fluphenazine) OR [compounds] (Haloperidol) OR [compounds] (Olanzapine) OR [compounds] (Quetiapine) OR [compounds] (Risperidone) OR [compounds] (Thiothixene) OR [compounds] (Ziprasidone)”.

Identification of genes associated with cognitive function in schizophrenia by GeneCards Database

The GeneCards database21 was used to identify genes associated with cognitive function in schizophrenia (step 3 of Fig 1). [publications] (schizophrenia) AND [disorders] (Schizophrenia AND [publications] (cognitive) AND [disorders] (cognitive) OR [functions] (cognitive) OR [pathways] (cognitive) is the keyword for searching in this step.

Identification of genes associated with cognitive function in schizophrenia by WebGestalt Database

The gene associated with cognitive function in schizophrenia was identified by the WebGestalt Database (http://www.webgestalt.org/) (step 4 of Fig 1), which stands for WEB-based GEne SeT AnaLysis Toolkit.22 Over-representation analysis (ORA) is a method used in this study. The upload gene list is collected from the previous step (step 2 of Fig 1), which is the identification of gene targets for antipsychotics.

Identification of genes associated with antipsychotics and cognitive function in schizophrenia

This step is a consequence of step 2 (identification of gene targets for selected antipsychotics) and step 3 (identification of genes associated with cognitive function in schizophrenia by GeneCards Database). Gene findings from the previous step were condition formatting, then filtered by matching in both terms of antipsychotics and cognitive function in schizophrenia. The step-by-step data condition formatting and filtering data in Microsoft Excel is used to find common genes between two gene lists. The gene identification starts by selecting the data column > Conditional formatting in the Home ribbon > Highlight cell rule > Duplicate Values, and then pressing OK. Then, select the data column > Sort & Filter in the home ribbon > Filter to apply the filter dropdown icon to selected columns.

Protein-protein interactions network analysis

The gene list from the previous step (step 5 of Fig 1) was further analyzed to identify protein-protein interactions and construct a protein-protein interactions network using the STRING database. The step to identify protein-protein interactions of the gene of interest starts with inputting the name of the gene of interest into the multiple proteins search box and then selecting auto-detect organism and search. The input of the Gene list might appear in several organisms. The Homo sapiens organism was selected from the list. Many proteins in Homo sapiens might similar to the input data. The matching protein was selected. A protein-protein interactions network was generated and visualized only connected nodes in interaction evidence mode, using three interaction sources: experiment, database and co-expression. The default setting was applied for the minimum required interaction score and the maximum number of interactions to show in the network display mode.

Network clustering

The network was clustered using the Markov Clustering

method (MCL), with three inflation parameters to establish the most interconnected cluster of the protein-protein interaction (PPI) network.23

Identification of a common gene associated with antipsychotics and cognitive function in schizophrenia from deference two methods

Gene findings from the WebGestalt and network clustering were condition formatting, then filtered by matching genes from the two lists. The condition formatting and filtering data settings in Microsoft Excel were used to find common genes between two gene lists.

Identification pathway of potential proteins involved in antipsychotics and cognitive function in schizophrenia The results from the previous step of deference

two methods, which involved the identification of a common gene associated with antipsychotics and cognitive function in schizophrenia, were used to identify the related pathway by the STRING database. The results of the related pathway were shown in functional enrichment in the analysis section.24

Comprehensive graphical user interface snapshots, input datasets, parameter configurations, filtering criteria, threshold values, and detailed procedural steps for each analysis are provided in the Supplemental Data.

RESULTS

The nine commonly used antipsychotics were chosen, including Aripiprazole, Chlorpromazine, Fluphenazine, Haloperidol, Olanzapine, Quetiapine, Risperidone, Thiothixene, and Ziprasidone.17-20,24

The results of gene targets for antipsychotics showed a total of 223 target genes (Fig 2). The results obtained from the GeneCards Database showed that 122 genes were associated with cognitive function in schizophrenia (Fig 2). The results from WebGestalt database analysis showed that 223 gene targets for nine selected antipsychotics were characterized into three main categories, including biological process, cellular component, and molecular function. Apart from 223 genes, the results showed that 31 genes were found to be associated with cognitive function in schizophrenia (Fig 2).

The 28 candidate genes associated with antipsychotics and cognitive function in schizophrenia were constructed into protein-protein interaction networks analysis using the STRING database. The results showed 28 Nodes of protein and 33 Edges, representing interactions between pairs of proteins. The protein-protein interaction was identified based on experimental, database, or coexpression evidence. Moreover, nine proteins, including CHAT,

Fig 2. Flow diagram illustrating the summary of results.

APOE, GAD1, PLA2G6, SLC18A2, IL10, MTHFR, COMT,

and TH, have no interaction with any protein (Fig 3A). The network clustering of 28 candidate genes associated with antipsychotics and cognitive function in schizophrenia was constructed using the STRING database. The results exhibited four network clusters, and the largest cluster comprises 12 proteins. The 7 more proteins were clustered in one cluster of 3 proteins and two clusters of 2 proteins, respectively. The nine disconnected nodes were hidden from the network clustering (Fig 3B). The 12 proteins of the large clustering

were selected for analysis in the next experiment.

The target gene associated with antipsychotics and cognitive function in schizophrenia, collected from 31

genes of the WebGestalt database analysis and 12 genes of network clustering, was taken into consideration using deference two methods. The six common genes associated with antipsychotics and cognitive function in schizophrenia were obtained from deference two methods, which include CHRM1, HRH3, DRD2, GRM5, HTR2A and SLC6A1. The gene symbol, description and antipsychotics interact with the 6 candidate genes are shown in Table 1.

The six common genes associated with antipsychotics and cognitive function in schizophrenia were further analyzed using KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis. The KEGG analysis results in 4 pathways, including neuroactive ligand-receptor

B

A

Fig 3. (A) Protein-protein interactions (PPI) network of proteins associated with antipsychotics and cognitive function in schizophrenia. The protein-protein interaction is characterized by three different types of evidence, including experimental, database, or coexpression. Each evidence of protein-protein interaction is shown in a different color of the connecting line. The experimental, database or coexpression evidence is indicated in purple, blue and black lines, respectively. (B) Protein-protein interactions network of proteins associated with antipsychotics and cognitive function in schizophrenia after network clustering and hiding disconnected nodes. The network clustering is indicated in color.

(A larger version of this figure is available in the Supplemental Data file.)

TABLE 1. Gene symbol, description, and list of antipsychotics interact with the common gene associated with antipsychotics and cognitive function in schizophrenia from deference two methods.

interaction, gap junction, calcium signaling pathway, and cAMP signaling pathway (Table 2). The observed gene count, background gene count, false discovery rate and matching proteins in the network of each pathway are shown in Table 2.

DISCUSSION

Recently, several lines of evidence emphasized the functional role of antipsychotic drugs in the cognitive function of psychiatric disorders, including schizophrenia.25-30,61-63 However, the mode of action and molecular mechanisms of antipsychotic drug-related cognitive function are not clearly understood. Bioinformatics is one of the tools that reduces the costs and time to study drugs and target-specific drugs.28,29 Moreover, network analysis is a bioinformatics approach that is employed to gain a broad and in-depth picture of protein-protein interaction (PPI) networks and signaling pathways.30

In the present study, nine frequently used antipsychotics were selected from the literature review. The GeneCards database was used to identify the drug target gene and gene of interest involved in cognitive function in schizophrenia. The next step was to analyze the gene lists with two different tools: WebGestalt Database, a functional enrichment analysis web tool, and STRING database, a web tool for studying protein-protein interactions. The overlapping six genes from the two previous steps may play an essential role in cognitive function. The potential gene list consists of 6 genes including CHRM1 (cholinergic receptor muscarinic 1), HRH3 (histamine receptor H3), DRD2 (dopamine receptor D2), GRM5 (glutamate metabotropic receptor 5), HTR2A (5-hydroxytryptamine receptor 2A) and SLC6A1 (solute carrier family 6 member 1). These genes were analyzed to find the related pathway using the KEGG database. Functional enrichment in KEGG

reveals 6 target proteins involved in several pathways, which play an important role in cognitive function, including neuroactive ligand-receptor interaction31-36, gap junction37, calcium signaling pathway38-40, and the cAMP signaling pathway.41

CHRM1 (Cholinergic Receptor Muscarinic 1) is a G protein-coupled receptor involved in acetylcholine binding and is implicated in cognitive function. Reduced cortical CHRM1 levels have been linked to cognitive decline and executive dysfunction in schizophrenia.42-45 HRH3 (Histamine Receptor H3) modulates histamine signaling in the CNS and regulates neurotransmitter release; H3 receptor antagonists show potential in treating cognitive impairment in schizophrenia without extrapyramidal side effects.42,46,47 DRD2 (Dopamine Receptor D2) participates in locomotion, reward, memory, and learning; mutations and polymorphisms in DRD2 are associated with schizophrenia risk and cognitive deficits.42,48,49 GRM5 (Glutamate Metabotropic Receptor 5), a glutamate receptor modulating synaptic plasticity, is linked to cognitive impairments in schizophrenia and may be a promising target to reverse these deficits.42,50,51 HTR2A (5-Hydroxytryptamine Receptor 2A) regulates neural activity and cognition; serotonergic modulation via 5-HT receptors contributes to cognitive and motor function improvements in schizophrenia treatment.42,52,53 SLC6A1 (Solute Carrier Family 6 Member 1) encodes a GABA transporter affecting synaptic inhibition, with impairments contributing to working memory deficits in schizophrenia.42,54

Nine antipsychotics bind to their protein target

with different action mechanisms, and most of them are receptor antagonists. It has been noted that the large group of proteins involves antipsychotics and cognitive function in schizophrenia are receptors of the

TABLE 2. List of pathways that contain potential proteins involved in antipsychotics and cognitive function in schizophrenia

neuroactive ligand-receptor interaction pathway, including CHRM1 (Cholinergic Receptor Muscarinic 1), DRD2 (Dopamine Receptor D2), GRM5 (Glutamate Metabotropic Receptor 5), HRH3 (Histamine Receptor H3), and HTR2A (5-Hydroxytryptamine Receptor 2A). These receptors are G-protein-coupled receptors (GPCRs).55,56 (Fig 4) The downstream cascade of DRD2, HRH3, and CHRM1 is adenylate cyclase (AD), which activates Cyclic AMP (cAMP) and protein kinase A (PKA) in a cAMP signaling pathway.57,58 PKA-mediated molecule that connects the cAMP signaling pathway and the gap junction. PKA might act on a gap junction as an inhibitor.59 Gap junctions hold intercellular channels that allow direct communication and direct transfer of small molecules between the cytosolic of neighboring cells. The communication of these channels is essential for many biological processes, such as apoptosis and differentiation. Taken together, a cAMP-PKA signaling pathway is a crucial mediator of long-term memory formation.41 The finding of new targets of antipsychotics in the present study is consistent with the recent evidence that glutamate metabotropic receptor modulators and serotonin agonists have the potential to improve cognitive impairments in schizophrenia.60 The involvement of GPCRs and their associated signaling pathways, such as neuroactive ligand-receptor interaction, calcium signaling, and cAMP signaling, in cognitive processes and antipsychotic drug action is well-established in schizophrenia research.64 This study

confirms these known mechanisms; however, the scientific contribution lies in the integrative identification of six overlapping target genes (CHRM1, HRH3, DRD2, GRM5, HTR2A, and SLC6A1) through dual bioinformatics approaches, refining specific drug-gene interaction candidates relevant to cognitive dysfunction.

Distinctly, the analysis highlights potential unique drug-gene interactions by mapping commonly prescribed antipsychotics to specific gene targets, some of which have been underexplored in population-specific contexts. For example, the inclusion of HRH3 and SLC6A1 as candidate genes suggests avenues beyond the classical dopaminergic and serotonergic targets, aligning with emerging therapeutic efforts to modulate histaminergic and GABAergic systems in schizophrenia cognition.65 Regarding the broad nature of the enriched pathways identified, this study advances understanding by contextualizing the role of these pathways specifically in cognitive impairment among schizophrenia patients. For instance, the cAMP-PKA signaling cascade identified here is crucial for synaptic plasticity and long-term memory formation, directly linking molecular targets to known cognitive deficits in schizophrenia.66 Furthermore, the neuroactive ligand-receptor pathway’s multiple GPCR targets illustrate how antipsychotics modulate neurotransmission

collectively affecting cognitive domains.

Comparatively, prior bioinformatics and network pharmacology studies in schizophrenia have predominantly

Fig 4. Schematic representation of antipsychotics’ action on receptors and signal transduction pathways.

focused on genetic susceptibility or broad molecular networks without integration of specific antipsychotic drug targets.67 This study contributes by bridging drug pharmacology and cognitive genomics, utilizing a combinatorial approach of gene target overlap, protein interaction networks, and pathway enrichment. Therefore, it provides a more targeted framework for identifying therapeutic candidates that could be further investigated in population-specific and pharmacogenetic contexts.64 Current evidence mostly reveals the role of antipsychotic action on psychotic symptoms and cognitive functions in schizophrenia patients.63,68,69 However, the mechanism of action of antipsychotic drugs on GPCRs related to the neurological processes of cognitive function is limited. The results of this study highlight the interaction of the molecular mechanisms of antipsychotic drugs on the neurological processes associated with psychosis and cognitive function. Therefore, this crucial evidence might suggest the conceptual framework for further study on pharmacological and clinical applications of novel antipsychotic drugs with dual effects on psychotic

symptoms and cognitive functions in schizophrenia.

CONCLUSION

The findings of this study support existing evidence that G-protein-coupled receptors (GPCRs) are critical targets through which antipsychotics exert their effects on cognitive function in schizophrenia. While the identification of enriched pathways such as neuroactive ligand-receptor interaction, calcium signaling, and cAMP signaling aligns with well-established biological mechanisms, this study refines these pathways by linking specific gene targets implicated in cognitive dysfunction in schizophrenia. Such integrative bioinformatics analyses enhance understanding of molecular interactions underlying cognitive deficits and provide a more precise molecular framework that connects antipsychotic pharmacodynamics with cognition-related genetic factors. This may facilitate the identification of candidate targets for future research aiming at developing therapies to address cognitive impairments in schizophrenia.

Data Availability Statement

Data will be made available on request.

DECLARATIONS

Grant and Funding Information

This research project is supported by National Research Council of Thailand (NRCT) and Mahidol University (grant number N42A650340 to B.C.).

Conflict of Interest

The authors declare no conflicts of interest related to this article.

Registration Number of Clinical Trial

None.

Author Contributions

Conceptualization and methodology, K.T., B.C., V.C., J.C.; Investigation, K.T.; Formal analysis, K.T., B.C., V.C.; Visualization and writing-original draft, B.C., J.C.; Writing-review and editing, B.C., J.C.; Supervision, B.C., V.C., V.S., J.C. All authors have read and agreed to the final version of the manuscript.

Use of Artificial Intelligence

None.

REFERENCES

1. Harvey PD, Lombardi J, Leibman M, White L, Parrella M, Powchik P, et al. Cognitive impairment and negative symptoms in geriatric chronic schizophrenic patients: a follow-up study. Schizophr Res. 1996;22(3):223-31.

2. Hughes C, Kumari V, Soni W, Das M, Binneman B, Drozd S, et al. Longitudinal study of symptoms and cognitive function in chronic schizophrenia. Schizophr Res. 2003;59(2-3):137-46.

3. Wu JQ, Chen DC, Tan YL, Xiu MH, De Yang F, Soares JC, et al. Cognitive impairments in first-episode drug-naive and chronic medicated schizophrenia: MATRICS consensus cognitive battery in a Chinese Han population. Psychiatry Research. 2016;238:196-202.

4. Seidman LJ, Buka SL, Goldstein JM, Tsuang MT. Intellectual decline in schizophrenia: evidence from a prospective birth cohort 28-year follow-up study. J Clin Exp Neuropsychol. 2006;28(2): 225-42.

5. Friedman JI, Harvey PD, Coleman T, Moriarty PJ, Bowie C, Parrella M, et al. Six-year follow-up study of cognitive and functional status across the lifespan in schizophrenia: a comparison with Alzheimer's disease and normal aging. Am J Psychiatry. 2001;158(9):1441-8.

6. Goldberg TE, Ragland JD, Torrey EF, Gold JM, Bigelow LB, Weinberger DR. Neuropsychological assessment of monozygotic twins discordant for schizophrenia. Arch Gen Psychiatry. 1990; 47(11):1066-72.

7. Riley EM, McGovern D, Mockler D, Doku VC, S OC, Fannon DG, et al. Neuropsychological functioning in first-episode psychosis--evidence of specific deficits. Schizophr Res. 2000; 43(1):47-55.

8. Bilder RM, Lipschutz-Broch L, Reiter G, Geisler SH, Mayerhoff DI, Lieberman JA. Intellectual deficits in first-episode schizophrenia: evidence for progressive deterioration. Schizophr Bull. 1992; 18(3):437-48.

9. Green MF. What are the functional consequences of neurocognitive deficits in schizophrenia? Am J Psychiatry. 1996;153(3):321-30.

10. Sharma T. Quetiapine-efficacy in different domains. Eur Neuropsychopharmacol. 2001;11 Suppl 4:S385-90.

11. Harvey PD, Lombardi J, Kincaid MM, Parrella M, White L,

    
    

    Powchik P, et al. Cognitive functioning in chronically hospitalized schizophrenic patients: age-related changes and age disorientation as a predictor of impairment. Schizophr Res. 1995;17(1):15-24.

12. Sharma T, Antonova L. Cognitive function in schizophrenia: Deficits, functional consequences, and future treatment. Psychiatr Clin North Am. 2003;26(1):25-40.

13. Faber G, Smid HG, Van Gool AR, Wiersma D, Van Den Bosch RJ. The effects of guided discontinuation of antipsychotics on neurocognition in first onset psychosis. Eur Psychiatry. 2012; 27(4):275-80.

14. Knowles EE, David AS, Reichenberg A. Processing speed deficits in schizophrenia: reexamining the evidence. Am J Psychiatry. 2010;167(7):828-35.

15. Elie D, Poirier M, Chianetta J, Durand M, Gregoire C, Grignon S. Cognitive effects of antipsychotic dosage and polypharmacy: a study with the BACS in patients with schizophrenia and schizoaffective disorder. J Psychopharmacol. 2010;24(7):1037-44.

16. Green MF, Marder SR, Glynn SM, McGurk SR, Wirshing WC, Wirshing DA, et al. The neurocognitive effects of low-dose haloperidol: a two-year comparison with risperidone. Biol Psychiatry. 2002;51(12):972-8.

17. Janicak PG, Marder SR, Tandon R, Goldman M. Schizophrenia recent advances in diagnosis and treatment. New York, NY: Springer; 2014.

18. John MECfCD, Communications S. AHRQ Comparative effectiveness reviews antipsychotic medicines for treating schizophrenia and bipolar disorder: A review of the research for adults and caregivers. Comparative Effectiveness Review Summary Guides for Consumers. Rockville (MD): Agency for Healthcare Research and Quality (US); 2005.

19. Chokhawala K, Stevens L. Antipsychotic Medications. StatPearls. Treasure Island (FL): StatPearls Publishing LLC.; 2019.

20. Andreasen NC, Pressler M, Nopoulos P, Miller D, Ho B-C. Antipsychotic Dose Equivalents and Dose-Years: A Standardized Method for Comparing Exposure to Different Drugs. Biol Psychiatry. 2010;67(3):255-62.

21. GeneCards - Human Genes | Gene Database | Gene Search [cited 2020 8 Feb]. Available from: https://www.genecards. org/.

22. Webgestalt(Web-based gene set analysis toolkit) [cited 2020 Feb 08]. Available from: http://www.webgestalt.org/.

23. STRING: functional protein association networks [cited 2020 Feb 8]. Available from: https://string-db.org/cgi/input.pl.

24. Jenraumjit R, Tethanyawarakool V, Kongkhamboot H, Piyatrakul

    N. Prescribing Pattern of Antipsychotic Long Acting Injections for the Treatment of Schizophrenia Patients in Suan Prung Psychiatric Hospital. J Psychiatr Assoc Thailand. 2018;63(4):335-48.

25. Sharma T. Cognitive effects of conventional and atypical antipsychotics in schizophrenia. Br J Psychiatry Suppl. 1999; 174(38):44-51.

26. Sakurai H, Bies RR, Stroup ST, Keefe RSE, Rajji TK, Suzuki T, et al. Dopamine D2 receptor occupancy and cognition in schizophrenia: analysis of the CATIE data. Schizophr Bull. 2013;39(3):564-74.

27. Knafo S, Venero C. Cognitive Enhancement: Pharmacologic, Environmental and Genetic Factors: Academic Press; 2014.

28. Dibyajyoti S, Bin ET, Swati P. Bioinformatics: The effects on the cost of drug discovery. Galle Medical Journal. 2013;18(1):

    44-50.

29. Buchan NS, Rajpal DK, Webster Y, Alatorre C, Gudivada RC, Zheng C, et al. The role of translational bioinformatics in drug discovery. Drug Discov Today. 2011;16(9-10):426-34.

30. Wu J, Vallenius T, Ovaska K, Westermarck J, Mäkelä TP, Hautaniemi S. Integrated network analysis platform for protein-protein interactions. Nat Methods. 2009;6(1):75-7.

31. Bolton JL, Marioni RE, Deary IJ, Harris SE, Stewart MC, Murray GD, et al. Association between polymorphisms of the dopamine receptor D2 and catechol-o-methyl transferase genes and cognitive function. Behav Genet. 2010;40(5):630-8.

32. Wess J. Muscarinic acetylcholine receptor knockout mice: novel phenotypes and clinical implications. Annu Rev Pharmacol Toxicol. 2004;44:423-50.

33. Barco A, Bailey CH, Kandel ER. Common molecular mechanisms in explicit and implicit memory. J Neurochem. 2006;97(6):1520-33.

34. Ménard C, Quirion R. Group 1 metabotropic glutamate receptor function and its regulation of learning and memory in the aging brain. Front Pharmacol. 2012;3:182.

35. Sadek B, Saad A, Sadeq A, Jalal F, Stark H. Histamine H3 receptor as a potential target for cognitive symptoms in neuropsychiatric diseases. Behav Brain Res. 2016;312:415-30.

36. Zhang G, Stackman Jr RW. The role of serotonin 5-HT2A receptors in memory and cognition. Front Pharmacol. 2015;6:225.

37. Mitterauer BJ. Possible role of glia in cognitive impairment in schizophrenia. CNS Neurosci Ther. 2011;17(5):333-44.

38. Kawamoto EM, Vivar C, Camandola S. Physiology and pathology of calcium signaling in the brain. Front Pharmacol. 2012;3:61.

39. Suwanjang W, Phansuwan-Pujito P, Govitrapong P, Chetsawang B. Calpastatin reduces calpain and caspase activation in methamphetamine-induced toxicity in human neuroblastoma SH-SY5Y cultured cells. Neurosci Lett. 2012;526(1):49-53.

40. Joshi AU, Kornfeld OS, Mochly-Rosen D. The entangled ER-mitochondrial axis as a potential therapeutic strategy in neurodegeneration: a tangled duo unchained. Cell Calcium. 2016;60(3):218-34.

41. Titus D, Furones C, Kang Y, Atkins CM. Age-dependent alterations in cAMP signaling contribute to synaptic plasticity deficits following traumatic brain injury. Neuroscience. 2013; 231:182-94.

42. Stelzer G, Rosen N, Plaschkes I, Zimmerman S, Twik M, Fishilevich S, et al. The GeneCards suite: from gene data mining to disease genome sequence analyses. Curr Protoc Bioinformatics. 2016;54:1.30.1-1.30.33.

43. Liao D-L, Hong C-J, Chen H-M, Chen Y-E, Lee S-M, Chang C-Y, et al. Association of muscarinic m1 receptor genetic polymorphisms with psychiatric symptoms and cognitive function in schizophrenic patients. Neuropsychobiology. 2003;48(2):72-6.

44. Dean B, Soulby A, Evin GM, Scarr E. Levels of [3H] pirenzepine binding in Brodmann's area 6 from subjects with schizophrenia is not associated with changes in the transcription factor SP1 or BACE1. Schizophr Res. 2008;106(2-3):229-36.

45. Scarr E, Sundram S, Deljo A, Cowie TF, Gibbons AS, Juzva S, et al. Muscarinic M1 receptor sequence: preliminary studies on its effects on cognition and expression. Schizophr Res. 2012; 138(1):94-8.

46. Zhang M, Ballard ME, Pan L, Roberts S, Faghih R, Cowart M, et al. Lack of cataleptogenic potentiation with non-imidazole H3 receptor antagonists reveals potential drug–drug interactions

    
    

    between imidazole-based H3 receptor antagonists and antipsychotic drugs. Brain Res. 2005;1045(1-2):142-9.

47. Jin C-Y, Anichtchik O, Panula P. Altered histamine H3 receptor radioligand binding in post-mortem brain samples from subjects with psychiatric diseases. Br J Pharmacol. 2009;157(1):118-29.

48. Cohen O, Weickert T, Hess J, Paish L, McCoy S, Rothmond D, et al. A splicing-regulatory polymorphism in DRD2 disrupts ZRANB2 binding, impairs cognitive functioning and increases risk for schizophrenia in six Han Chinese samples. Mol Psychiatry. 2016;21(7):975-82.

49. Nkam I, Ramoz N, Breton F, Mallet J, Gorwood P, Dubertret C. Impact of DRD2/ANKK1 and COMT polymorphisms on attention and cognitive functions in schizophrenia. PloS One. 2017;12(1):e0170147.

50. Matosin N, Newell KA, Quidé Y, Andrews JL, Teroganova N, Green MJ, et al. Effects of common GRM5 genetic variants on cognition, hippocampal volume and mGluR5 protein levels in schizophrenia. Brain Imaging Behav. 2018;12(2):509-17.

51. Moghaddam B. Targeting metabotropic glutamate receptors for treatment of the cognitive symptoms of schizophrenia. Psychopharmacology (Berl). 2004;174(1):39-44.

52. Shimizu S, Mizuguchi Y, Ohno Y. Improving the treatment of schizophrenia: role of 5-HT receptors in modulating cognitive and extrapyramidal motor functions. CNS Neurol Disord Drug Targets. 2013;12(6):861-9.

53. Blasi G, De Virgilio C, Papazacharias A, Taurisano P, Gelao B, Fazio L, et al. Converging evidence for the association of functional genetic variation in the serotonin receptor 2a gene with prefrontal function and olanzapine treatment. JAMA Psychiatry. 2013;70(9):921-30.

54. Lewis DA, Moghaddam B. Cognitive dysfunction in schizophrenia: convergence of γ-aminobutyric acid and glutamate alterations. Arch Neurol. 2006;63(10):1372-6.

55. Janušonis S. Functional associations among G protein-coupled neurotransmitter receptors in the human brain. BMC Neurosci. 2014;15:16.

56. Regard JB, Sato IT, Coughlin SR. Anatomical profiling of G protein-coupled receptor expression. Cell. 2008;135(3):561-71.

57. Sassone-Corsi P. The cyclic AMP pathway. Cold Spring Harb Perspect Biol. 2012;4(12):a011148.

58. Yan K, Gao LN, Cui YL, Zhang Y, Zhou X. The cyclic AMP

    signaling pathway: Exploring targets for successful drug discovery. Mol Med Rep. 2016;13(5):3715-23.

59. Pidoux G, Taskén K. Anchored PKA as a gatekeeper for gap junctions. Commun Integr Biol. 2015;8(4):e1057361.

60. Miyamoto A, Mikoshiba K. Probes for manipulating and monitoring IP3. Cell Calcium. 2017;64:57-64.

61. Feber L, Peter NL, Chiocchia V, Schneider-Thoma J, Siafis S, Bighelli I, et al. Antipsychotic Drugs and Cognitive Function: A Systematic Review and Network Meta-Analysis. JAMA Psychiatry. 2025;82(1):47-56.

62. Lee M, Cernvall M, Borg J, Plaven-Sigray P, Larsson C, Erhardt S, et al. Cognitive Function and Variability in Antipsychotic Drug–Naive First-Episode Psychosis. JAMA Psychiatry. 2024; 81(5):468-76.

63. Kelebie M, Kibralew G, Tadesse G, Rtbey G, Aderaw M, Endeshaw W, et al. Effectiveness of antipsychotic medication in patients with schizophrenia in a real world retrospective observational study in Ethiopia. Sci Rep. 2025 Feb 7.

64. Nnadi CU, Malhotra AK. Individualizing Antipsychotic Drug Therapy in Schizophrenia: The Promise of Pharmacogenetics. Curr Psychiatry Rep. 2005;7(1):1-7.

65. Begni V, Marchesin A, Riva MA. IUPHAR review-Novel therapeutic targets for schizophrenia treatment: A Translational perspective. Pharmacol Res. 2025;214:107690.

66. Titus DJ, Furones C, Kang Y, Atkins CM. Age-dependent alterations in cAMP signaling contribute to synaptic plasticity deficits following traumatic brain injury. Neuroscience. 2013; 231:182-94.

67. Faustmann TJ, Corvace F, Faustmann PM, Ismail FS. Influence of antipsychotic drugs on microglia-mediated inflammatory responses in schizophrenia. Front Psychiatry. 2025;16:1522128.

68. Pitanupong J, Karakate A, Tepsuan L, Sritrangnant G. Attitudes toward long-acting injectable antipsychotics among schizophrenia patients in southern Thailand: A multihospital-based cross-sectional survey. Siriraj Med J. 2022;74(3):193-201.

69. Ueapanjasin P, Thavornwattanayong W, Lertsirimunkong J, Chaiyakittisopon K.Cost-effectiveness analysis of long-acting injectable once-monthly of aripiprazole compared with long-acting injectable once-monthly paliperidone palmitate for the treatment of stable schizophrenia patients in Thailand. Siriraj Med J. 2023;75(10):725-35.

‌Sexual Dimorphism in Cranial and Post-Cranial Skeletal Elements: Forensic Implications for Sex Estimation in a Contemporary Thai Population

Napakorn Sangchay, M.D., Ph.D.1, Kavin Tangmanpakdeepong, B.Sc.1,3, Saidontree Boonyarud, B.Sc.1,3, Sujinthara Wansopha, B.Sc.1,3, Natipong Chatthai, M.Sc.1,2, Patara Rattanachet, Ph.D.1, Jirapa Chetsawang, M.D., Ph.D.1,*

1Department of Anatomy, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand, 2College of Medical Science, Western University,

Bangkok, Thailand, 3Mahidol University International College, Mahidol University, Nakhon Pathom, Thailand.

*Corresponding Author: Jirapa Chetsawang E-mail: napakorn.sac@mahidol.ac.th

Received 25 August 2025 Revised 17 September 2025 Accepted 20 September 2025 ORCID ID:http://orcid.org/0000-0002-7776-6456 https://doi.org/10.33192/smj.v77i12.277229

All material is licensed under terms of the Creative Commons Attribution 4.0 International (CC-BY-NC-ND 4.0) license unless otherwise stated.

ABSTRACT

Objective: The analysis of human skeletal remains is instrumental in forensic and anthropological contexts, particularly for establishing biological profiles of unidentified individuals. Determining sex via skeletal examination is a fundamental component of this process and traditionally involves morphological assessment and metric analysis of pelvic and cranial bones. Nevertheless, the precision and reliability of these methodologies—whether through morphological evaluation or morphometric analysis—remain subjects of ongoing debate and scrutiny. This research set forth to investigate the efficacy of morphological and morphometric analysis in sex estimation by focusing on cranial and post-cranial long bones within contemporary Thai population.

Materials and Methods: The study sample comprised 204 skulls (105 from males, and 99 from females) and 200 sets

of long bones of upper (humerus, radius, and ulna) and lower extremities (femur and tibia). Multiple measurements were systematically obtained from various anatomical regions of each bone, and measurements of extremity long bones were compared between the left and right sides.

Results: The analysis revealed statistically significant differences in these metrics between sexes, indicating the

potential utility of this approach for sex classification. However, despite achieving high levels of accuracy, the studied methodology yielded some classification errors, which suggests some potential limitations.

Conclusion: The findings suggest that both inherent skeletal variability in cranial and post-cranial morphology

within contemporary Thai population and the specific analytical techniques employed can markedly influence the accuracy of sex determination. These factors harbor and confer important implications for forensic and anthropological applications.

Keywords: Sexual dimorphism; cranial and post-cranial skeletal elements; forensic implications for sex estimation; contemporary Thai population (Siriraj Med J 2025; 77: 858-876)

INTRODUCTION

In forensic anthropology, constructing a biological profile for unidentified human remains is crucial for identification, including sex, age, stature, and ancestry estimations. These parameters are essential for matching ante-mortem records and other data. Notably, a swift and organized skeletal analysis for sex determination is vital for refining subsequent biological profile estimations.1 Skeletal sex estimation is a critical step that requires

both expertise and precise decision-making due to its foundational role in the identification process. Sexual dimorphism in skeletal elements results from genetic and hormonal influences, which contribute to distinct traits differentiating male and female skeletons. To enhance accuracy and minimize subjectivity, sex assessments should be conducted and documented meticulously, thereby strengthening the identification process. The distinct skeletal traits influenced by genetic and hormonal factors differentiate males and females, underscoring the need for reliable and less subjective assessment techniques.2 In forensic investigations, accurate sex assessment from skeletal remains is essential for human identification. The pelvis and cranium are preferred for analysis due to their pronounced sexual dimorphism, enabling precise sex estimation. The os coxae is particularly favored for its marked dimorphism and high accuracy. However,

when these elements are absent or severely fragmented, alternative bones must be considered. Human long bones, or postcranial elements, serve as a secondary option due to their sexual dimorphism, though morphometric methods using long bones are limited, especially in databases focusing on Asian genetic and ethnic backgrounds.3,4

The pelvis and skull are traditionally considered the most reliable skeletal components for sex determination due to well-defined dimorphism.5 Accurate assessment requires complete or intact structures, often compromised in forensic settings due to taphonomic degradation or recovery limitations. In such cases, long bones are valuable substitutes due to their sexually dimorphic features influenced by musculoskeletal development and hormonal regulation. However, they lack specific morphological traits indicative of sex, limiting their reliability for sex estimation.6

Numerous studies have examined the use of the human skull and long bones of the extremities for sex estimation. However, a critical limitation of many such investigations is the lack of intra- and inter-observer error assessments, which impedes a thorough evaluation of the reliability of the methodology being studied or the methodologies being compared.7 Cranial visual assessment of or for sexually dimorphic traits offers a rapid and practical method for sex estimation, but this

technique is highly subjective and less reproducible. This approach, particularly when evaluating the skull, involves analyzing varying morphological characteristics to distinguish between sexes, focusing on features such as the mastoid process, nuchal crest, and orbital traits.8 In addition to pelvic morphology, features in long bones, such as femoral head diameter, can indicate biological sex and are suitable for morphometric analysis. This method is effective with complete or nearly complete skeletal elements, but its reliability diminishes with fragmented remains due to taphonomic damage. Thus, exploring alternative methodologies, like digital imaging techniques to analyze long bone characteristics, becomes

necessary.9,10

While gross morphological analysis of the skull offers acceptable accuracy for sex estimation, it also has some associated limitations as follows. This technique is susceptible to intra- and inter-observer variability, leading to inconsistent assessments. The reliance on subjective judgment complicates its utility, particularly in skeletal specimens with ambiguous features. To enhance precision and reliability, direct pelvic morphometric measurements are proposed as a more refined and reliable approach.11,12 Metric analysis of the skull provides greater precision

in sex determination, often surpassing 90% accuracy, compared to visual morphological assessments. However, this method requires an intact or nearly complete skull to accurately assess landmarks. The mastoid process, in particular, demonstrates high reliability, with a sex determination success rate of 96.5%, supported by studies reporting approximately 90% accuracy from measurements in this region.13

Both non-metric and metric cranial analyses depend on identifiable skeletal features, which necessitate well-preserved bones. Current methodologies primarily derive from skeletal collections of specific populations, such as Black and White Americans, with limited data for Asian and other groups.14 Researchers underscore the critical need for population-specific databases to improve sex estimation accuracy in diverse populations.15

Asymmetry between the bones on the left and right sides is attributed to handedness. Approximately 90% of the population is right-handed. Handedness influences bone morphology, with the dominant limb generally exhibiting larger bones. Forensic studies also show that the right side, including hand bones, tends to be larger, particularly in right-handed individuals.16 Larger bones in the right hand are more common in right-handed individuals, with this difference being more pronounced in males. However, similar correlations are less consistent for left-handed individuals.17

This study evaluates sexual dimorphism in cranial and post-cranial elements for forensic sex estimation, assessing inter-observer reliability and error rates using gross morphological and metric approaches. Conducted in contemporary Thai population with dry skulls and extremity long bones, this study explores the method’s applicability in forensic contexts. Additionally, morphometric analysis of the bilateral long bones examines the impact of asymmetry on sex estimation, considering the influence of sexual dimorphism.

MATERIALS AND METHODS

The study utilized human bone samples from the Siriraj Anatomical and Anthropological Bone Centre (Si-AABRC), of the Department of Anatomy, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand. The protocol for this study was exempted from Siriraj Institutional Review Board (SIRB) approval.

Cranial sexual dimorphism and morphometric measurements

Gross morphological assessment was conducted for each bone, focusing on traits such as the mastoid process, nuchal crest, and cranial contour, to estimate sex. Each cranium underwent detailed visual evaluation of specific dimensions and landmarks, including the supraorbital ridges, opisthocranion, and mastoid process. A sex classification was made if at least three traits indicated a particular sex. The supraorbital ridges were assessed for size and prominence, the opisthocranion for its most posterior point, and the mastoid process for size and robustness. All skulls were sequentially numbered and segregated to ensure unbiased subsequent craniometric analysis.

Following morphological assessment, cranial morphometric measurements were performed, as detailed in Table 1. An independent re-evaluation by two blinded investigators was conducted to enhance accuracy and reduce inter-observer variability. The craniometric parameters assessed are illustrated in Fig 1, with both observers unaware of prior measurements and the biological sex of each evaluated skull.

Post-cranial morphometric measurements

The upper extremity long bones included the humerus, radius, and ulna, on both sides, and the lower extremity bones included the femur and tibia, on both sides. Bones with deformities, fractures, osteophytes, or prosthetic modifications were excluded. Measurements were made using an osteometric board and digital vernier calliper. Post-cranial measurements are summarized in Table 2.

TABLE 1. Description of cranial landmarks and morphometric measurements.

Cranial Length (Front to Back) Measurement of distance from glabella to opisthocranion

Cranial Index (Cranial Width/Cranial Length) x 100

Mastoid Height Measurement of porion to end of mastoid process

Inter-supraorbital Foramen Distance Measurement of left supraorbital foramen or notch to right

Fig 1. Cranial parameters used in cranial measurement method. All measurements were performed using the same digital vernier calliper and the same spreading calliper.

Statistical analyses

Cranial skeletal elements:

Prior to parameter measurements and analysis, two independent observers measured over 200 dry crania, yielding 204 viable samples. The study compared cranial measurements between sexes and evaluated classification accuracy via visual assessments measurement reliability. Statistical comparisons used either the independent t-test or Mann-Whitney U test, depending on data normality,

to identify significant sexual dimorphism. The primary goal was to identify cranial features with pronounced sexual dimorphism.

A contingency table assessed classification accuracy, comparing predicted sex with actual sex. Metrics such as sensitivity, specificity, precision, and overall accuracy evaluated the sex determination method. Measurement reliability was assessed with Cronbach’s alpha, and intraclass correlation coefficients (ICC) were calculated for normally

TABLE 2. Description of post-cranial long bone landmarks and morphometric measurements.

Maximum Anatomical Length U-MAL Distance between the proximal end of olecranon process to distal

end of styloid process

Maximum Olecranon Process Width MOPW Distance between maximum of olecranon process from medial to lateral side

TABLE 2. Description of post-cranial long bone landmarks and morphometric measurements. (Continue)

distributed variables. ICC estimates the proportion of variance due to true differences, with reliability ranges, ensuring measurement consistency for valid comparisons.

Post-cranial skeletal elements:

The independent t-test compared two groups, with statistical significance set at p<0.001 (α=0.05). Paired t-tests evaluated intra-individual differences between the left and right sides. Discriminant function analysis assessed sex prediction accuracy, selecting functions with the highest classification performance and significant variables.

Observation error and reliability analysis

Inter-observer measurement error was evaluated. Two observers independently measured the skeletal samples: the primary investigator and a faculty member of the Department of Anatomy with 10 years of experience. Each observer assessed all parameters blinded to the demographic data. The goal was to quantify measurement error and repeatability, as analyzed by SPSS Statistics software (SPSS, Inc., Chicago, IL. USA). Repeatability was evaluated using technical error of measurement (TEM), relative TEM (rTEM), and coefficient of reliability (R). An R-value approaching 1 indicates high agreement.18

RESULTS

Descriptive statistics for all measurements are summarized in Tables 3-5, including means and standard deviations. All measurements showed significant differences between males and females (all p<0.05).

Cranial skeletal elements

Table 3 provides descriptive statistics for cranial variables by sex. The independent t-test revealed that males have significantly larger cranial length, width, cranial index (Fig 2-A&B), cranial base length (Fig 2C), mastoid length (Fig 2D), inter-orbital foramen distance (Fig 2E), and inter-infraorbital foramen distance (Fig 2G). The Mann-Whitney U test additionally showed that the inter-supraorbital foramen distance (Fig 2F) was significantly larger in males (p=0.005), indicating strong sexual dimorphism suitable for sex estimation.

Diagnostic accuracy of visual assessment

Table 4 summarizes classification results: 75 male skulls correctly identified, 17 female skulls misclassified as male, 30 male skulls misclassified as female, and 82 female skulls correctly classified. The method’s sensitivity was 71.43% (95% confidence interval [95% CI]: 61.79–79.82),

and specificity was 82.83% (95% CI: 73.94–89.67), with a

TABLE 3. Descriptive statistics of cranial landmarks and morphometric variables (a Independent t-test, b Mann-Whitney U test).

a Independent t-test, b Mann-Whitney U test

Fig 2. Comparative analysis of craniometric parameters. The figure displays a box and whisker plot comparing the means of craniometric parameters compared between male (n=105) and female (n=99) crania, namely: (A) Cranial length and width.

(B) Cranial Index. (C) Cranial base length. (D) Mastoid length. (E) Inter-orbital foramen length. (F) Inter-supraorbital foramen distance. (G) Inter-infraorbital foramen distance. Significance levels (p<0.01) are indicated for each parameter.

TABLE 4. Contingency table of sex classification and diagnostic accuracy by visual assessment and reliability of cranial morphometric assessment.

TABLE 5. Descriptive statistics of post-cranial landmarks and morphometric variables.

positive likelihood ratio of 4.16 and a negative likelihood ratio of 0.34. The positive predictive value (PPV) was 81.52%, negative predictive value (NPV) was 73.21%, and overall accuracy was 76.96%.

Extremity long bones

Descriptive statistics (Table 5) and boxplots (Fig 3) show significant sex differences (p<0.001) across all humeral, radial, and ulnar parameters. Table 5 displays descriptive statistics for femoral and tibial measurements. Independent t-tests (Fig 5) confirmed significant sex differences for all parameters (p<0.001).

Bilateral comparisons

Table 5 presents bilateral measurements. Paired t-tests (Fig 4) showed significant symmetry, except for HuSND, HuHMD, HuEB, HuOFD, RaDRW and UlMAL

in females; and, HuHMAL, HuSND, RaRMD, RaDRB, RaDRW and UlUMAL in males.

Paired t-tests (Fig 6) showed significant asymmetry, except for Fe-MD (p=0.847 in females, and p=0.817 in

males). In tibiae, males demonstrated greater variability, with females showing differences in T-MD (p<0.001), T-PMW (p=0.008), T-PLW (p=0.022), while males differed significantly in five parameters, all p<0.01, as illustrated in Fig 6.

Reliability analysis: Technical error of measurement

The coefficient of reliability (R) for both cranial and post-cranial measurements is shown in Tables 4 and 6, respectively.

Reliability of craniometric measurements

Table 4 reports inter-observer reliability. Cronbach’s alpha indicated high internal consistency, with parameters such as cranial length, inter-orbital distance, and inter-infraorbital distances all exceeding 0.90. Other metrics, including cranial width, cranial index, cranial base length, mastoid length, and supraorbital distances, showed good reliability (α 0.75-0.90), confirming measurement consistency.

Reliability of osteometric measurements of the extremity long bones

Table 6 illustrates the R-values for this interobserver reliability analysis. Most upper and lower extremity long bone morphometrics show a reliability above the cut-off point, (high to very high). However, two tibial morphometrics exhibit a reliability below the cut-off point, including TPLW (r=0.64) and DTW (r=0.64).

DISCUSSION

Sex estimation from skeletal remains can be achieved

A C E

B D F

Fig 3. Boxplot comparing the results of the independent t-test between males and females. (A) Right side of humerus. (B) Left side of humerus. (C) Right side of radius. (D) Left side of radius. (E) Right side of ulna. (F) Left side of ulna.

A

C

E

F

B D

Fig 4. Boxplot comparing the results of the paired samples t-test between the left and right sides. (A) Humerus in males. (B) Humerus in females. (C) Radius in males. (D) Radius in females. (E) Ulna in males. (F) Ulna in females. (Red box indicates p<0.001).

A

C

B

D

Fig 5. Boxplot comparing the results of the independent t-test between males and females. (A) Right side of femur. (B) Left side of femur.

(C) Right side of tibia. (D) Left side of tibia. (***p<0.001).

A

B

D

C

Fig 6. Boxplot comparing the results of the paired samples t-test between the left and right sides. (A) Femur in males. (B) Femur in females.

(C) Tibia in males. (D) Tibia in females. (*p<0.05, **p<0.01, and ***p<0.001).

through the visual assessment of morphological variations stemming from sexual dimorphism. Key indicators include skeletal size and robustness, which are influenced by extrinsic factors such as biomechanical loads, as well as intrinsic factors like genetic makeup and hormonal influences. Both cranial and postcranial skeletal elements exhibit distinguishable sexual dimorphism in adult individuals. Among the available methodologies, combined non-metric and metric analysis of skull morphology is considered the second most reliable approach for sex determination in adult skeletal remains, following pelvic morphology assessment.19 Within the cranium, features of the mastoid region are regarded as particularly informative. Morphological characteristics of the nuchal crest and mental eminence form the basis of cranial morphological sex estimation techniques, which are recognized for their high accuracy. Subsequent validation studies have reinforced and refined the reliability of skull morphological traits, leading to enhanced predictive accuracy and decreased classification errors.20,21

The findings of this study demonstrate that cranial

measurements obtained by multiple observers, following anthropometric standards, exhibit high levels of reliability. Cronbach’s alpha coefficients ranged from 0.832 to

0.971, indicating strong to excellent internal consistency between/among observers’ measurements. These results are consistent with prior research, which recommended a reliability threshold of R>0.95 to ensure measurement dependability and accuracy. Notably, cranial length, inter-orbital foramen distance, and inter-infraorbital foramen distance showed the highest reliability coefficients at 0.971, 0.920, and 0.908, respectively. The elevated reliability scores of these parameters suggest that they are among the most sexually dimorphic features in the cranium. Furthermore, these results support the premise that craniometric measurements, when conducted following standardized protocols and independent assessments, can be both consistent and reproducible across different observers, thereby serving as robust tools for forensic sex determination. Overall, the results endorse the reliability and reproducibility of craniometric methods in forensic contexts.

The statistical analyses, employing independent

t-tests and Mann-Whitney U tests, revealed statistically significant differences in various cranial measurements between males and females. Consistent with existing literature on the Thai population, male crania demonstrated larger mean measurements than females.22,23 The high

TABLE 6. Assessment of the coefficients of reliability for post-cranial extremity long bones measurements.

reliability scores further reinforce the statistical validity of these findings. This concordance with prior studies underscores the efficacy of craniometric techniques in accurately sexing skulls and supports their application in forensic investigations. Although visual assessment yielded an accuracy of approximately 77%, with sensitivity and specificity of 71.43% and 82.83%, respectively, it remains inherently subjective and susceptible to inter-observer variability. This aligns with earlier research highlighting the potential for inaccuracies and examiner dependence associated with visual methods. In contrast, craniometric techniques offer objective, quantifiable data, reducing the likelihood of subjective bias. While direct comparison of the accuracy between these methods was limited by differing evaluation criteria, the superior reliability of craniometric measurement underscores its potential as an efficient and objective approach for forensic sex estimation.24,25

Themorphologies of the mastoid process, orbital region,

and nuchal crest are key indicators of sexual dimorphism. Additionally, other cranial features have been identified as potential sex determinants. These anatomical areas are notably resilient, often withstanding destructive forces. Employing morphometric measurements combined with stepwise discriminant functions generally achieves more accurate sex classification than visual assessments alone. However, this study found that morphometric analysis can be less accurate in sex differentiation compared to direct observational methods.

Sexual dimorphism and morphometric analysis of extremity long bone

Sexual dimorphism in the femur and tibia is strongly influenced by both sex hormones and biomechanical factors, leading to generally larger skeletal dimensions in males compared to females. This study revealed statistically significant differences between sexes across all measured femoral and tibial parameters (p<0.001), with males exhibiting greater mean values. The observed dimorphism is largely attributed to the influence of androgens and estrogens, particularly testosterone, which promote bone growth and increased bone mass in males, especially during puberty. This hormonal effect contributes to the development of more pronounced skeletal features, including key landmarks on the femur and tibia. Specifically, the femoral measurements in this study support findings from previous research in Thai population, showing similar values for midshaft diameter (Fe-MD), head diameter (Fe-HD), and neck diameter (Fe-ND).26,27 However, the current study found that the maximum bone length of the femur (Fe-MBL) is notably

greater, suggesting potential population-specific growth trends or methodological differences.

When comparing femoral bone lengths, including the maximum length (Fe-MBL) and anatomical length (Fe-MAL), the values observed in this study are smaller than those reported in Sri Lankan and Indian populations, yet larger than those found in Japanese population.28 In comparison with an Ancient Anatolian population, the measurements for females are closely aligned, whereas male measurements are noticeably smaller. For midshaft diameter (Fe-MD), this study’s results are slightly greater than those reported in Indian population and generally consistent with Sri Lankan data, although the Sri Lankan female values are slightly smaller.29 Notably, in Sri Lankan population, male and female midshaft diameters are nearly identical, highlighting minimal dimorphism in that metric. Compared to ancient Anatolian population, midshaft diameters in this study are again closely aligned in females, but marginally smaller in males, suggesting possible population-specific variation or environmental influences on skeletal development.

On the distal femur (Fe-MCW and Fe-LCW), the

measurements of this study were found to be slightly smaller in size compared to Korean population.30,31 Importantly, it must be noted there is no comparative analysis of Fe-BB due to a difference in measurement methods since other studies used epicondylar breadth – the maximum length between medialmost to lateral most points on the epicondyle of the femur – instead of the definition used in this study.

Similarly, the tibial measurements in this study demonstrate marked sexual dimorphism, aligning with findings from previous research. When compared to earlier data from Thai population, the maximum tibial length (T-MBL) and distal tibia breadth (DTB) in this study are slightly smaller, while the proximal tibial measurements—specifically the medial and lateral condylar widths (TPMW and TPLW)—are slightly larger. The tibial plateau breadth (TPB), however, remains comparable between studies. In comparison with Sri Lankan population, this study reports a notably smaller anatomical length (T-MAL), while TPB measurements are larger.32 Additionally, TPMW and TPLW values are slightly greater than those observed in Japanese and Brazilian populations, suggesting region-specific variation in proximal tibial morphology.33 Compared to ancient Anatolian population, T-MBL values are marginally larger in females and slightly smaller in males, further highlighting population and sex-related differences in tibial dimensions.34

The midshaft diameter (T-MD) of the tibia in the

present study was found to be notably larger than that reported in the Indian population. When compared with ancient Anatolian population, female measurements were distinctly larger, while overall values were generally comparable. It is important to acknowledge, however, that methodological variations exist across studies, particularly in the anatomical landmarks used for measurement—many researchers take the diameter at the level of the nutrient foramen, which may influence reported values. Regarding the distal tibia, the breadth of the inferior articular surface (TPB) closely aligns with data from Sri Lankan population but is significantly larger than values recorded in Japanese population.35-37 Similarly, the width of the inferior articular surface (DTW) in this study was found to be markedly greater than that of Kenyan population, highlighting inter-population variability in distal tibial morphology.38

From all of the aforementioned points, the femur

and tibia can both be regarded as effective and reliable indicators for sex determination, and the measurement methods used for both the femur and tibia can be used in the process of sex determination.

TEM and reliability

It is well-recognized that observer error arising from visual or metric variables can result in inconsistencies in evaluating sexual dimorphism and determining sex. Recent studies indicate that visual assessment is prone to substantial inter-observer error due to vague variable definitions, heavy reliance on the observer’s prior experience, and the seriation process used to categorize individuals. This error analysis demonstrates that geometric morphometrics achieves high levels of intra- and inter-observer agreement.

The results of the technical error of measurement (TEM) analysis, which is used to assess inter-assessor reliability, indicate variable levels of agreement across different skeletal parameters. For the humerus, six out of eight parameters—namely HMAL, HHD, AND, HMD, EB, and CD—demonstrated very high reliability, whereas SND and OFD exhibited only high reliability scores. In the case of the radius, half of the parameters, including RMAL, RHD, and DRW, showed very high interpretative consistency, whereas RND, RMD, and DRB were characterized as having only high reliability. For the ulna, all parameters—UMAL, MOPW, MCPW, and UMD—exhibited high reliability. This variation can be attributed to differences in the anatomical aspects of each bone, such as the specific landmarks used for measurement, including SND and OFD in the humerus, RND, RMD, and RBD in the radius, and UMAL, MOPW, MCPW,

and UMD in the ulna. The instances where parameters yielded only high rather than very high reliability may stem from variability in landmark identification, often influenced by factors such as the absence of precise measurement landmarks—particularly in regions like the surgical neck of the humerus, where measurement relies heavily on the assessor’s judgment.

Values reflect measurement consistency for both femoral and tibial parameters, except for the maximum bone length measurements of the femur (Fe-MBL) and tibia (T-MBL), due to the absence of comparable data. Using an R-value threshold of ≥0.90, most femoral measurements exceeded this cutoff, indicating high interobserver reproducibility, except for Fe-MCW and Fe-LCW. In contrast, most tibial measurements did not reach this threshold with only two parameters—T-MAL and DTB—demonstrating high reliability. These results suggest that femoral morphometric measurements generally exhibit greater interobserver consistency, whereas tibial measurements show variable reliability, with only selected parameters achieving robust repeatability.

Diagrams were used to clarify the definitions of each parameter during measurements; however, despite standardization, measurement errors remained possible due to limitations in observer interpretation. A key source of error is the variation in observer experience. In this study, one observer had no prior experience in skeletal measurement, whereas the other was an expert in forensic anthropology, which likely contributed to discrepancies, particularly in the identification of anatomical landmarks. Variations in understanding and interpreting the diagrams may have further contributed to minor differences in landmark annotation. Additionally, the morphology of the landmarks themselves influences measurement reliability. Many tibial measurements, along with two femoral parameters (Fe-MCW and Fe-LCW), involve landmarks with curved or rounded contours, which can be ambiguous and challenging to delineate precisely, thereby reducing measurement accuracy.

Cranial morphologies can often be consistently

distinguished even by observers without prior experience. However, morphologies that are not clearly defined present significant challenges for consistent determination. Our findings indicate that even experienced observers analyzing skulls for sexual dimorphism may inconsistently interpret coordinate landmarks for sex assessment. While the robustness of the mastoid process and the pronounced nuchal crest are strong indicators of male sex, morphological trait evaluation still leads to varying levels of prediction accuracy between/among observers. This suggests that anatomical landmarks on cranial

bones may be misinterpreted. The description of these landmarks varies across the literature. It is crucial to describe quantitative methodologies that utilize cranial features exhibiting sexual dimorphism since they are pertinent to forensic practitioners. Additionally, understanding the impact of cranial sexual dimorphism on classification accuracy in sex estimation is essential.39-41

The results of the reliability analysis of craniometric measurements taken by two inter-observers (the intraclass correlation coefficient [ICC], specifically Cronbach’s alpha), was used to assess the internal consistency and reliability of the measurements across various cranial parameters. That analysis revealed that the craniometric measurements exhibit high internal consistency, with all parameters showing good to excellent reliability. When interpreted according to the ranges established by Koo and Li, the Cronbach’s alpha values for cranial length, inter-orbital foramen distance, and inter-infraorbital foramen distance indicate excellent reliability (α>0.90).42 Other parameters, including the cranial width, cranial index, cranial base length, mastoid length, and inter-supraorbital foramen distance showed good reliability (0.75 ≤ α ≤ 0.90). These high reliability scores suggest that the measurements are consistent and repeatable across different observers.43,44

Furthermore, the definition of specific landmarks

measured on the cranium and long bones of the extremities, such as the tibia, presents a recurring challenge in certain regions. To ensure measurement accuracy, all assessments must be both reproducible and independent. Consequently, future research should include evaluations of interobserver error in these measurements to minimize the risk of sex misclassification.

Effects of skeletal asymmetry and siding comparison The paired samples t-test results of the upper extremity long bones indicate that two of the eight parameters—HMAL and SND—exhibit statistically significant differences between the left and right sides in males, with p-values less than 0.001. Additionally, analysis of the radius bones revealed three parameters—RMD, DRB, and DRW—that demonstrate significant bilateral differences at p-values below 0.001. Conversely, examination of the ulna shows that only UMAL exhibits significant asymmetry with a p-value under 0.001. The variation in the degree of bilateral differences, ranging from high to very high significance, is likely influenced by factors, such as handedness, which predominantly affects bone size disparity between sides. Supporting this, Walters et al. (1998) reported that right-handed individuals tend to have larger bones in their dominant

hand, highlighting the functional role of dominance in skeletal asymmetry.45-47

Humans, as bipeds, typically distribute body weight evenly across the lower limbs during upright stance. Minimal bilateral differences in femoral and tibial morphometrics are, therefore, generally expected, so many previous studies have therefore analyzed only one limb for sex estimation. However, our findings revealed significant asymmetry between sides for both femoral and tibial measurements, which can be attributed to factors such as differential biomechanical loading, environmental influences, habitual limb preferences, and localized mechanical stress - all of which can induce side-specific bone remodeling. Degenerative joint conditions, notably osteoarthritis affecting the hip and knee, may further contribute to asymmetries, with parameters, such as femoral head diameter, bicondylar breadth, and proximal tibial measurements, showing consistent side differences. In the femur, almost all parameters demonstrated significant asymmetry with p-values below 0.05, which is consistent with prior studies that reported greater measurements on the left side, particularly in the lower limbs.48,49

Similarly, most femoral measurements in our study

were larger on the left side in both sexes. Carvallo and Retamal (2020) studied a Chilean population and found bilateral symmetry predominantly in proximal femoral parameters, but they did not explore the underlying causes.50 In the tibia, asymmetry was also present, but less pronounced, with males showing greater asymmetry than females. These findings align with van der Gaast, et al. (2022), who identified asymmetry in the tibial plateau potentially related to previous injuries, while the distal tibia exhibited high symmetry, consistent with the findings of Verbakel, et al. (2024). Nonetheless, caution in interpretation is warranted due to the relatively limited sample size.51,52

Given the observed inherent asymmetry, measurements should be taken bilaterally to ensure accuracy - especially for the femur. Despite the tibia generally demonstrating greater symmetry than the femur, bilateral assessment of the tibia remains advisable. Due to the possibility of unpredictable laterality in fragmentary remains at scene investigations, it is prudent to prioritize measurements of parameters known for bilateral symmetry when only partial femoral or tibial fragments are available.

Although most adult skeletons display characteristic sexual dimorphism, the accuracy of sex estimation is influenced by several factors, including population variability, age, and pathological or taphonomic changes. The degree and expression of sexual dimorphism can vary significantly both within and across different populations. Therefore,

it is crucial to employ population-specific data when applying these techniques in forensic contexts. The extent of sexual dimorphism and the differences between sexes differ across populations, and both morphological and metric methods are used for sex estimation. Morphological attributes—such as shape, specific traits, and relative size differences—are key indicators. Techniques focusing on pelvic shape, measurements, and the presence or absence of distinctive pelvic features are generally preferred since these tend to demonstrate greater sexual dimorphism and higher accuracy. Other morphological traits may also suggest sex, but they are often less reliable than those with pronounced dimorphic features. Employing appropriate instrumentation, standardized protocols, advanced analytical software, and integrating multiple measurements through multivariate approaches can improve the reliability of sex assessments, although individual measurements can still yield reasonably accurate results.

Sample size is a crucial consideration since the validation

of classification functions necessitates cross-population studies to assess their robustness and generalizability. The use of approximately 400 individuals in this study may affect prediction accuracy and elevate the risk of technical errors. Expanding the sample size could mitigate these issues and enhance the overall effectiveness of the method.

Lastly, understanding the factors that influence prediction accuracy and classification errors—particularly in cranial and post-cranial sex estimation—is vital, given that the cranium is primarily governed by hormonal rather than mechanical factors - unlike the extremity long bones. This makes it especially useful in cases where the population of origin from which an unidentified skeleton originates from is unknown. Future research should focus on collecting and analyzing cooperative datasets from diverse populations, both for pelvic parameters and other skeletal features that display sexual dimorphism, to refine and validate forensic sex estimation methods across different population groups.

Limitations and suggestions

Future research should include measurements of additional osteometric landmarks, such as cranial landmark distances of cranium and fibular morphometrics of post-cranial bones and assess both intra- and interobserver reliability across all measurement methods.

Larger regionally diverse, and ethnically varied samples are recommended to evaluate the reliability, repeatability, and accuracy of sex estimation methods. The resulting data may facilitate the development of

discriminant functions and help to estimate other biological profiles, such as stature.

The fact that our sample size of 204 crania is slightly lower than Yamane’s suggested 214 likely exerted minimal impact on the results of our study. However, variation in observer expertise significantly influenced measurement accuracy in our study. The author’s limited experience may have affected the accuracy of visual assessment, and despite efforts to reduce variability, differences in examiner skill and physical and mental condition contribute to inter-observer variability. Standardized training and calibration are, therefore, necessary to enhance consistency.

Outliers in descriptive statistics, particularly for the inter-supraorbital foramen distance (p=0.056), can be attributed to population-specific variations, such as the presence of multiple foramina or notches. These variations may have affected measurement accuracy, highlighting the need for further research to account for such differences in diverse populations.

CONCLUSION

Determining sex from skeletal remains is a fundamental component of human identification in forensic investigations. Although the skull is regarded as the second most preferable skeletal element for sex estimation, it is essential to account for measurement errors, which can impact the accuracy of sex determination. The adoption of alternative morphometric approaches, such as direct measurement of long bones of the extremities, may offer a viable alternative with superior accuracy—particularly when the pelvis or skull are not available. It is crucial to identify and understand the instrumental and methodological factors that influence accurate sex prediction. The selection of an appropriate analytical approach is paramount since it significantly influences the overall process of forensic human identification and helps reduce the likelihood of misclassification.

Data Availability Statement

The data used in this study was obtained from the Siriraj Anatomical and Anthropological Bone Research Centre (Si-AABRC) database, Department of Anatomy, Faculty of Medicine Siriraj Hospital, Mahidol university, Bangkok, Thailand. Access to this data was approved through Si-AABRC data access procedures. The data is not publicly available but can be requested from the Si-AABRC with appropriate permissions.

ACKNOWLEDGEMENT

None.

DECLARATIONS

Grants and Funding Information

None.

Conflicts of Interest

None.

Registration Number of Clinical Trial

Not Applicable.

Author Contributions

Conceptualization and methodology, N.S. and J.C.; Investigation, K.T., S.B., S.W., N.C. and J.C.; Formal

analysis, K.T., S.B., S.W., N.C. and P.R.; Visualization and writing – original draft, K.T., S.B., S.W., N.C. and P.R.; Writing – review and editing, N.S.; Supervision,

N.S. All authors have read and agreed to the final version of the manuscript.

Use of Artificial Intelligence

No artificial intelligence was used in this study.

Ethics/IRB Approval

SIRB Protocol No. 195/2564 (Date of Proof: 22 March 2021) and 680/2564 (Date of Proof: 25 August 2021).

REFERENCES

1. Wang X, Liu G, Wu Q, Zheng Y, Song F, Li Y. Sex estimation techniques based on skulls in forensic anthropology: a scoping review. Egypt J Forensic Sci. 2024;19(12):e0311762.

2. Lye R, Min H, Dowling J, Obertová Z, Estai M, Bachtiar NA, et al. Deep learning versus human assessors: forensic sex estimation from three-dimensional computed tomography scans. Sci Rep. 2024;14:30136.

3. Duangto P, Mahakkanukrauh P. Sex estimation from upper limb bones in a Thai population. Anat Cell Biol. 2020;53(1):36–43.

4. Krishan K, Chatterjee PM, Kanchan T, Kaur S, Baryah N, Singh RK. A review of sex estimation techniques during examination of skeletal remains in forensic anthropology casework. Forensic Sci Int. 2016;261:165.e1–8.

5. Selliah P, Martino F, Cummaudo M, Indra L, Biehler-Gomez L, Campobasso CP, et al. Sex estimation in middle and late adulthood: reliability of pelvic morphological traits and long bone metrics on an Italian skeletal collection. Forensic Sci Int. 2020; 134(5):1683-690

6. Blanc M, Knecht S, Nguyen K, Poulain C, Quatrehomme G, Alunni V, et al. Sexual dimorphism of the humerus bones in a French sample: comparison of several statistical models including machine learning models. Int J Legal Med. 2025;139(3):1395–408.

7. Houston SK, Brits D, Myburgh J, Liebenberg L. The impact of age-related changes in the skull on sex estimation using morphoscopic traits. Int J Legal Med. 2025.

8. Lewis CJ, Garvin HM. Reliability of the Walker cranial nonmetric method and implications for sex estimation. J

   Forensic Sci. 2016;61(3):743-51.

9. Wysocka J, Cieslik A, Danel D. Sex estimation using measurements of the proximal femur in a historical population from Poland. Anthropological Review. 2023;86(1):37-49.

10. Fliss B, Lüthi M, Fürnstahl P, Christensen AM, Sibold K, Thali M, et al. CT-based sex estimation on human femora using statistical shape modeling. Am J Phys Anthropol. 2019;169(2): 279–86.

11. Setiawati R, Rahardjo P, Ruriana I, Guglielmi G. Anthropometric study using three-dimension pelvic CT scan in sex determination among adult Indonesian population. Forensic Sci Med Pathol. 2022;19(1):24-33.

12. Torimitsu S, Makino Y, Saitoh H, SakumaA, Ishii N, Yajima D, et al. Morphometric analysis of sex differences in contemporary Japanese pelves using multidetector computed tomography. Forensic Sci Int. 2015;257:530.e.1-7.

13. Houston SK, Brits D, Myburgh J, Liebenberg L. The impact of age-related changes in the skull on sex estimation using morphoscopic traits. Int J Legal Med. 2025.

14. Tallman SD. Cranial nonmetric sexual dimorphism and sex estimation in East and Southeast Asian individuals. Forensic Anthropology. 2019;2(4):204–21.

15. Lye R, Obertová Z, Bachtiar NA, Franklin D. Validating the use of clinical MSCT scans for cranial nonmetric sex estimation in a contemporary Indonesian population. Int J Legal Med. 2024;138(4):1559–71.

16. Pérez-Criado L, Rosas A, Bastir M, Pastor JF. Humeral laterality in modern humans and Neanderthals: a 3D geometric morphometric analysis. Anthropol Sci. 2017;125(3):117–28.

17. Kumar S, Voracek M, Singh M. The effects of hand preference and sex on right–left asymmetry in dorsal digit lengths among adults and children. Early Hum Dev. 2021;153:105293.

18. Langley NR, Jantz LM, McNulty S, Maijanen H, Ousley SD, Jantz RL. Error quantification of osteometric data in forensic anthropology. Forensic Sci Int. 2018;287:183-9.

19. Sangchay, N, Dzetkuličová, V, Zuppello M, Chetsawang J. Consideration of Accuracy and Observational Error Analysis in Pelvic Sex Assessment: A Study in a Thai Cadaveric Human Population. Siriraj Med J. 2022;74(5):330–9.

20. Rogers TL. Sex determination of human remains through cranial morphology. J Forensic Sci. 2005;50(3):493-500.

21. Kruger GC, L’Abbe’ EN, Stull KE, Kenyhercz MW. Sexual dimorphism in cranial morphology among modern South Africans. Int J Legal Med. 2015;129(4):869–75.

22. Mahakkanukrauh P, Sinthubua A, Prasitwattanaseree S, Ruengdit S, Singsuwan P, Praneatpolgrang S, et al. Craniometric study for sex determination in a Thai population. Anat Cell Biol. 2015;48(4): 275–83.

23. Sinthubua A, Ruengdit S, Das S, Mahakkanukrauh P. A new method for sex estimation from maxillary suture length in a Thai population. Anat Cell Biol. 2017;50(4):261–4.

24. Saini V, Srivastava R, Rai RK, Shamal SN, Singh TB, Tripathi SK. Sex estimation from the mastoid process among North Indians. J Forensic Sci. 2012;57(2):434-9.

25. Ekizoglu O, Hocaoglu E, Inci E, Can IO, Solmaz D, Aksoy S, et al. Assessment of sex in a modern Turkish population using cranial anthropometric parameters. Leg Med (Tokyo). 2016;21:45-52.

26. Monum T, Prasitwattanaseree S, Das S, Siriphimolwat P, Mahakkanukrauh P. Sex estimation by femur in modern Thai population. Clin Ter. 2017;168(3):e203–e207.

    
    

27. Duangto P, Mahakkanukrauh P. Sex estimation from upper limb bones in a Thai population. Anat Cell Biol. 2020;53(1):36–43.

28. Ozer I, Katayama K. Sex determination using the femur in an ancient Japanese population. Coll Antropol. 2008;32(1):67-72.

29. Ranaweera L, Cabral E, Dissanayake DMPV, Lakshan WSV. Estimation of sex from the osteometric measurements of the femur in a contemporary Sri Lankan population. Int J Morphol. 2022;40(4):1009–17.

30. Cho H-J, Kwak D-S, Kim I-B. Morphometric evaluation of Korean femurs by geometric computation: comparisons of the sex and the population. BioMed Res Int. 2015;2015:730538.

31. Kim J-B, Lyu S-J, Kang H-W. Are Western knee designs dimensionally correct for Korean women? A morphometric study of resected femoral surfaces during primary total knee arthroplasty. Clin Orthop Surg. 2016;8(3):254–61.

32. Nanayakkara D, Vadysinghe AN, Nawarathna LS, Sampath H. Determination of sex from the tibia in a contemporary Sri Lankan population. J Forensic Sci Med. 2019;5(1):24–28.

33. Uehara K, Kadoya Y, Kobayashi A, Yamano Y. Anthropometry of the proximal tibia to design a total knee prosthesis for the Japanese population. J Arthroplasty. 2002;17(8):1028–32.

34. Özer BK, Özer İ, Sağır M, Güleç E. Sex determination using the tibia in an ancient Anatolian population. Mediterr Archaeol Archaeom. 2014;14(2):329–36.

35. Gupta C, Nayak N, Kalthur SG, D'Souza AS. A morphometric study of tibia and its nutrient foramen in South Indian population with its clinical implications. Saudi J Sports Med. 2015;15(3):244–8.

36. Nanayakkara D, Vadysinghe AN, Nawarathna LS, Sampath H. Determination of sex from the tibia in a contemporary Sri Lankan population. J Forensic Sci Med. 2019;5(1):24–28.

37. Tiwari A, Mahendru A, Priya A. An anatomical study of the tibia in the North Indianpopulation. Int J Hum Anat. 2019;2(1):1–7.

38. Misiani MK, Amuti T, Darbar S, Mandela P, Maranga E, Obimbo M. Sex determination from dimensions of distal tibiae in adult Kenyans: A discriminant function analysis. Translational Research in Anatomy. 2020;20:100075.

39. Torres HR, Morais P, Fritze A, Burkhardt W, Kaufmann M, Oliveira B, et al. Anthropometric landmarking for diagnosis of cranial deformities: Validation of an automatic approach and comparison with intra- and interobserver variability. Ann Biomed Eng. 2022;50(9):1022–37.

40. Kotěrová AP, Santos F, Bejdová S, Rmoutilová R, Attia MH, Habiba A, et al. Prioritizing a high posterior probability threshold leading to low error rate over high classification accuracy:

    the validity of MorphoPASSE software for cranial morphological sex estimation in a contemporary population. Int J Legal Med. 2024;138(2):1759–68.

41. Arigbabu OA, Liao IY, Abdullah N, Mohamad Noor MH. Computer vision methods for cranial sex estimation. IPSJ Trans Comput Vis Appl. 2017;9:19.

42. Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med. 2016;15(2):155–63.

43. Richard AH, Parks CL, Monson KL. Accuracy of standard craniometric measurements using multiple data formats. Forensic Sci Int. 2014;242:177–85.

44. Kizilgoz V, Aydin S, Aydemir H, Keles P, Kantarci M. Interobserver and intraobserver reliability of skull base angles measured on magnetic resonance images. World J Clin Cases. 2024;12(34): 6687–95.

45. Walters J, Koo WWK, Bush A, Hammami M. Effect of hand dominance on bone mass measurement in sedentary individuals. J Clin Densitom. 1998;1(4):359–67.

46. Nandi ME, Olabiyi O, Okubike EA, Cyprain IE. A study of bilateral asymmetry of upper extremity and its effects on stature reconstruction amongst Nigerians. Aus J Forensic Sci. 2018; 1(8):978–88.

47. Auerbach BM, Ruff CB. Limb bone bilateral asymmetry: variability and commonality among modern humans. J Hum Evol. 2006;50(2):203–18.

48. Marques S, Pinto C, Ferreira MT, Garcia S, Curate F. Sex Estimation from the Fibula and Tibia: A Study in Three Portuguese Reference Collections. Forensic Sci. 2023;5(1):2.

49. Mittino G, Langstaff H, García-Donas JG. Sex and Stature Estimation on the Tibia: A Virtual Pilot Study on a Contemporary Hispanic Population. J R Anthropol Inst. 2024;30(3):1–15.

50. Carvallo D, Retamal R. Sex estimation using the proximal end of the femur on a modern Chilean sample. Forensic Sci Int Rep. 2020;2:100077.

51. van der Gaast N, Dunning H, Huitema JM, Waters A, Jaarsma RL, Doornberg JN, et al. The symmetry of the left and right tibial plateau: a comparison of 200 tibial plateaus. Eur J Trauma Emerg Surg. 2023;49(1):69–74.

52. Verbakel J, Boot MR, van der Gaast N, Dunning H, Bakker M, Jaarsma RL, et al. Symmetry of the left and right tibial plafond; a comparison of 75 distal tibia pairs. Eur J Trauma Emerg Surg. 2024;50(6):2877–82.

‌Minimum Ischiopubic Ramus Width as a Single Metric for Sex Estimation: A Pilot Study

Robert W. Mann, Ph.D.1, Sittiporn Ruengdit, Ph.D.2,*, Patara Rattanachet, Ph.D.3,4, Napakorn Sanchay, M.D., Ph.D.3,4

1Department of Anatomy, Biochemistry, and Physiology, John A. Burns School of Medicine, The University of Hawai`i at Mānoa, Honolulu, Hawaii,

USA, 2Department of Forensic Medicine, Faculty of Medicine, Chiang Mai University, Chiang Mai, Thailand, 3Department of Anatomy, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand, 4Siriraj Anatomical and Anthropological Bone Research Centre, Department of Anatomy, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand.

*Corresponding author: Sittiporn Ruengdit E-mail: neu.sittiporn@gmail.com

Received 21 September 2025 Revised 2 November 2025 Accepted 9 November 2025 ORCID ID:http://orcid.org/0000-0002-3220-9146 https://doi.org/10.33192/smj.v77i12.277804

All material is licensed under terms of the Creative Commons Attribution 4.0 International (CC-BY-NC-ND 4.0) license unless otherwise stated.

INTRODUCTION

Estimating the biological sex of an individual through visual and metrical examination of the human skeleton is one of the four components of the biological profile, consisting of age at death, sex, ancestry, and stature. Establishing the biological profile of skeletal remains is important to medicolegal authorities as it provides them with information that may help narrow the possibilities or lead to the identification of a decedent. Reported accuracies in estimating sex of a complete skeleton range from 64 to 100%1, with the most reliable visual features being present in the pelvis, followed by the cranium.2 The pelvis, due to its sexually dimorphic size and shape features that are related to child birthing, can be used to reliably estimate sex in about 99% of individuals compared to 91% using the cranium. Constant re-examination of existing metrical and visual methods and technologies, and the development of new methods for estimating sex through examination of the human skeleton, continue to provide medicolegal analysts with higher accuracy rates3 for establishing sex in both complete and incomplete human skeletal remains.

In forensic investigations, complete skeletal remains

are not always recovered, requiring the use of skeletal analysis methods tailored to fragmented bones. The pelvis, being one of the skeletal structures most susceptible to fractures4, is often involved in these cases. Most sex estimation methods—whether traditional morphoscopic and morphometric techniques or shape analysis using geometric morphometrics—typically focus on the entire innominate bone or require a complete pubic

or ischiopubic complex.4 However, in archaeological contexts for example, the preservation of the pubic region rarely exceeds 30%.5 Additionally, the upper half of the ischiopubic ramus is more commonly preserved compared to the lower half. Therefore, the present study aims to assess the effectiveness of the minimum width of the ischiopubic ramus as a single measurement for sex estimation.

MATERIALS AND METHODS

This research protocol has been certified as exempt by the Research Ethics Committee of the Faculty of Medicine, Chiang Mai University (Exemption 0525/2025). All procedures were conducted in accordance with the ethical standards of the committee and with the Declaration of Helsinki. A total of 135 adult male and female left os coxae (innominate or hip bones) in the Mann-Labrash Osteology Collection at the John A. Burns School of Medicine University of Hawaii, Siriraj Anatomical and Anthropological Bone Research Centre at the Department of Anatomy at Siriraj Hospital School of Medicine, Mahidol University, and Osteology Research and Training Center in the Department of Anatomy at Chiang Mai University School of Medicine, Thailand were randomly selected and examined. The sample consisted of 73 adult males and 62 adult females of known age, sex, and identity donors in the Willed Body Program at the three universities. The Mann-Labrash Osteology Collection consists of known-identity individuals representing the population diversity of Hawaii who died between 1974 and 2020, including individuals of Japanese, Chinese,

Korean, Vietnamese, Hawaiian, African American, and European ancestry.6 The osteological collection curated at the Osteology Research and Training Center at Chiang Mai University School of Medicine consists of Thais, primarily from Northern Thailand, and who died between 2003 and 2014.6 The Siriraj Anatomical and Anthropological Bone Research Centre is composed of modern Thai and Chinese-descent individuals who donated their bodies for medical research between 1933 and 2020.7 Author RM measured the Hawaii sample, and SR measured the Northern Thai samples. PR and NS compiled data for the Siriraj samples. An additional holdout sample of 24 cases (12 males and 12 females) was selected to validate the efficiency of the obtained equation or model and its demarcation point.

The minimum mediolateral width of the ischiopubic ramus (minIPR) was obtained using a Mitutoyo SC-6”C sliding caliper by visually and metrically locating the narrowest mediolateral portion of the ramus, approximately 3 cm lateral to the most inferior part of the pubic symphysis. The narrowest mediolateral part of the IPR in many individuals was visible as a “pinched”, concave or biconcave area formed by the inferior margin of the IPR and the inferior margin of the obturator foramen in the area where the ischium and pubic bones join. The measurement was obtained while avoiding any bony projections, osteophytes, or evidence of trauma such as a fracture or healing callus. The minIPR was, however, rarely affected or altered by

localized or systemic bone disease such as cancer or osteoporosis. To obtain the minimum width of the IPR the blade-like portion of the arms of the caliper was held perpendicular to the long axis and ventral surface of the IPR (Fig 1) and moved slightly in all directions to obtain the narrowest mediolateral width. The measurement was rounded to the nearest millimeter. Each minIPR was measured without knowing the sex of an individual.

Data analysis

Statistical analyses were conducted using R software (version 4.5.1).8 Inter-observer error analysis was performed to evaluate their reliability and repeatability. Twenty cases, 10 males and 10 females, were randomly selected and remeasured by the third author (PR). All measurements recorded by the second author (SR), acting as the first observer, and the third author (PR), acting as the second observer, were compared using the mean absolute difference between repeated measurements, the mean difference expressed as a percentage of average bone size, as well as the technical error of measurement (TEM) and the coefficient of reliability (R).9-11 Descriptive statistics were computed to summarize the data. An independent samples t-test was performed to compare the minIPR variable between males and females, with statistical significance set at p < 0.05.

Numerous machine learning approaches have been applied in forensic anthropology and related disciplines. For this study, we selected the most commonly used

Fig 1a-b. Position of the sliding caliper when measuring the left ischiopubic ramus in an elderly female. (a) Note that the caliper is held perpendicular to the ventral surface of the ischiopubic ramus at its narrowest point to obtain the smallest measurement. It’s easier to take this measurement when holding the calipers upside down as in this image. (b) Slightly different angle showing the position of the caliper arms against the ischiopubic ramus.

predictive models, including seven classifiers: XGBoost (XGB), Support Vector Machine (SVM), Random Forest (RF), Logistic Regression (LR), Artificial Neural Network (ANN), Linear Discriminant Analysis (LDA), and Decision Tree (DT). The performance of both classical statistical methods and machine learning classifiers was then evaluated.

The dataset was partitioned into a training set (80%) for model construction and a test set (20%) for evaluating classification accuracy.

Model evaluation

The models’ performance was evaluated using accuracy, precision, recall, and F1-score metrics. Accuracy measures the overall agreement between the actual and predicted sex by dividing the number of correctly predicted instances by the total number of instances. In medical literature, precision—also known as predictive value—indicates the probability that a sex estimation is accurate based on the algorithm’s prediction. Recall, a group-specific performance metric, also known as sensitivity, represents the probability that the algorithm correctly estimates sex. Lastly, F1-score is a metric used to evaluate the performance of a classification model, especially in scenarios where the data is imbalanced. It is the harmonic mean of precision and recall, offering a balance between the two.

RESULTS

Table 1 presents the median and mean absolute differences, as well as the mean differences expressed as a percentage of the mean bone size, for the interobserver error analysis. Technical Error of Measurement (TEM) and the correlation coefficient (R) were also calculated to assess measurement repeatability. The obtained R value of 0.90 indicates an acceptable level of reliability for the minIPR measurement.11 Although the mean percentage difference was moderately high (9.59%), it was comparable to the TEM value (0.73 mm) when evaluated against the grand mean bone size, and notably lower than the mean difference in minIPR observed between males and females (2.7 mm).

Table 2 contains descriptive statistics including means and standard deviations for minIPR measurements of males and females. Independent t-test revealed the average measurements in males are statistically significantly wider than those in females (p < 0.05) with a mean difference of

2.7 mm, highlighting the presence of sexual dimorphism. The classification accuracy of all models ranged from 63.0% to 77.8%, with the Random Forest (RF) model showing the lowest accuracy, while Logistic Regression (LR) and Linear Discriminant Analysis (LDA) achieved the highest accuracy. Notably, LR and LDA also demonstrated superior performance in precision (71.4%), recall (83.3%), and F1 score (76.9%). In contrast,

TABLE 1. Interobserver error statistics for the minimum mediolateral width of the ischiopubic ramus.

Note: Median abs diff refers to Median absolute difference; Mean abs diff refers to Mean absolute difference; TEM refers to Technical error of measurement; %Mean diff refers to Mean difference as percentage of mean size.

TABLE 2. Descriptive statistics for the minimum mediolateral width of the ischiopubic ramus.

RF recorded the lowest values across these metrics. Regarding recall, the k-Nearest Neighbors (KNN) and Decision Tree (DT) models matched the highest values achieved by LR and LDA (Fig 2).

The Shapiro-Wilk test was used to assess normality, and the p-values were not statistically significant (p > 0.05), indicating that the data followed a normal distribution. The total correct classification accuracy obtained from both sexes was 77.8%. Logistic regression model fitting is tabulated in Table 3. The model using minIPR as a predictor variable is described by the following equation (females classified with negative values, whereas males are classified with positive values):

Sex = -4.236 + (0.354 × minIPR)

The model accurately classified the sex of 77.8% of individuals, achieving a sensitivity of 83.33% and a specificity of 73.33% with a demarking point of 11.977

mm. This demonstrates a strong discriminative ability, with an area under the curve (AUC) obtained from ROC curve of 0.85 and a Kappa value of 0.557. In the logistic regression model, a demarking point was established, indicating that a minIPR value of less than 11.977 mm suggests a female, while a value greater than this threshold suggests a male. Additionally, twenty-four holdout samples were randomly selected to evaluate the logistic regression model and its demarking point. The model demonstrated moderate performance, similar to that obtained from the LR model developed from the training set, achieving an accuracy of 75%.

DISCUSSION

The interobserver error statistics for the minIPR show relatively moderate to high error; however, in terms of millimeters, this error appears to be smaller than the sexual dimorphism observed in the minIPR

Fig 2. Comparison of classification model performance for sex estimation using the minimum ischiopubic ramus width. Seven statistical and machine learning classifiers—XGBoost, Support Vector Machine, Random Forest, Logistic Regression, Linear Discriminant Analysis, K-Nearest Neighbors, and Decision Tree—were evaluated based on Accuracy, Precision, Recall, and F1 Score.

TABLE 3. Coefficient obtained from logistic regression.

Hosmer and Lemeshow goodness of fit (GOF) test X-squared = 13.069, df = 8, p-value = 0.1095

measurements between males and females. Therefore, the potential measurement error is likely to have a minimal impact on sex estimation when using this measurement. Additionally, strictly adhering to the definition and carefully obtaining the absolute minimum IPR measurements by repeatedly measuring the areas visually assessed as the ‘narrowest’ could help reduce the potential for error.

The Phenice method has proven to be an accurate and reliable method for estimating sex in adult skeletons since it was developed. All three features of the Phenice method have been shown to yield high accuracy rates, particularly the ventral arc (96%)12-14, making the anterior pelvis the most reliable area in the skeleton for estimating sex.

A few studies have investigated sexual dimorphism using the minimum mediolateral width of the ischiopubic ramus15,16, which is in the anterior pelvic region. Dixit et al. reported high significance in Wilk’s lambda and F-ratio tests, consistent with Luo’s findings15, which achieved higher accuracy (81.7%) than the present study. The terminology and definition of the ischiopubic ramus employed in these studies closely matched that proposed in the present study. Dixit et al.16 refined Luo’s vague definition of “the minimum thickness of the ischiopubic ramus” by providing clearer measurement criteria, though neither study addressed interobserver measurement reliability.15 The definition described in Dixit’s study was “the least straight distance between the ischiopubic ramus and nearest obturator foramen margin”. Although both Luo and Dixit’s studies proposed a relatively similar definition of ischiopubic ramus width as compared to the present study, they did not report the classification accuracy for this measurement alone. All the high classification accuracies derived from their studies still required a combination of measurements. For instance, Luo’s study required four measurements, including “the minimum thickness of ischiopubic ramus” as they defined it, to obtain 81.7% accuracy. Similarly, Blake et al.17 included the ischiopubic ramus measurement among pubic measurement combinations analyzed using linear discriminant function analysis, achieving a cross-validated classification accuracy of 87.1%. However, like Luo and Dixit’s studies, the study did not specify the classification accuracy when using IPR alone. Ischiopubic ramus thickness (IPR), the term used in Blake et al.17, is a pubic measurement similar to minIPR. As mentioned in Blake et al17, the IPR was defined as “a distance between the inferior-most point of medial obturator foramen to the narrowest point inferior to the pubic symphysis” while the minIPR, used in the present study, is the minimum mediolateral width of the ischiopubic

ramus locating the narrowest mediolateral portion of the ramus, approximately 3 cm lateral to the most inferior part of the pubic symphysis. Although Luo, Dixit, and Blake’s studies reported higher classification accuracies (>80%) using similar definitions of ischiopubic ramus width, their methods relied on multiple measurements, whereas the present study demonstrates the utility of a single minimum ischiopubic ramus width (77.8%). Patriquin et al.18 proposed an LDA equation based on pubic measurements, including pubic bone width, with an accuracy of 69–79% for White and Black populations. However, this equation required measuring pubic bone height, and its definition of pubic bone width differed from the minIPR used in the present study.

Apart from traditional morphometric measurements, geometric morphometric analysis of the ischiopubic region has been explored.4,19,20 Black analyzed landmarks and semi-landmarks along the ischiopubic ramus, however, they did not clearly define or apply the narrowest area of the ischiopubic ramus measured mediolaterally as defined in the present study.4 Although Gonzalez and colleagues investigated the sexual dimorphism efficiency of the ischiopubic complex and achieved a high prediction accuracy of 93.4%, they used landmarks along the borders of the ischiopubic region.20 Bytheway and Ross19 also examined various os coxal shapes, including the ischiopubic ramus, which they defined as the narrowest point inferior to the pubic symphysis, using a geometric morphometric approach. However, their study did not report the sex estimation accuracy derived from the ischiopubic ramus shape alone.

Numerous studies have proposed multivariate and

univariate equations using pelvic measurements for sex estimation, achieving high classification accuracy (> 80%).21,22 However, most of these methods still require an almost intact pelvis. Similar to the ischiopubic ramus index, a promising proportion requiring only two measurements has been proposed for sex estimation from fragmented pelves.23,24 However, this index also necessitates an almost complete ischiopubic region. Additionally, the uncertainty in defining the acetabular point increases the potential for observer error with this index.23 Therefore, a single-metric approach, as proposed in the present study, offers an alternative for sex estimation from fragmented or incomplete pelves, despite its moderate classification accuracy (77.8%).

Table 4 presents a comparison of sex classification accuracies using a single measurement of the ischiopubic region across different studies. The classification accuracy of 77.8% achieved in the present study falls within the medium to high range compared to other studies. This

TABLE 4. A comparison of sex classification accuracies using a single measurement of the ischiopubic region across different studies.

accuracy is approaching 80%, which is an acceptable rate for sex estimation in forensic anthropology. Furthermore, it is nearly identical to the accuracy obtained from the maximum diameter of the femoral head, a widely used single measurement for sex estimation.25

In forensic investigations, complete skeletal remains are not always recovered, necessitating skeletal analysis methods for fragmented or incomplete bones. The pelvis is among the skeletal structures that are particularly prone to fractures.4 Most sex estimation methods focus on the entire innominate or require a complete pubic or ischiopubic complex.4,22,23,28 The pubic rami are among the most fractured structures in cases of moderate to severe trauma.29,30 Isolated fractures of a single ramus, particularly the ischial ramus, are also observed, with these fractures being more frequent in the elderly.31 Furthermore, the superior rami are more often fractured than the inferior rami. As a result, the minimum mediolateral width of the ischiopubic ramus, typically located in the upper part of the ischiopubic ramus about three centimeters below the inferior border of the pubic symphysis, shows promise as a useful measurement for sex estimation in fragmentary pelves. Future studies should empirically evaluate the reliability of the minIPR on experimentally fragmented or taphonomically altered os coxae. Such validation would provide critical evidence of the method’s robustness and practical applicability in forensic contexts where pelvic completeness is often compromised.

While this method can be used in addition to the

Phenice method, its strength as an indicator of sex is when

dealing with highly fragmented or incomplete skeletal remains. Essentially, if a 2-to-3-centimeter portion of the ischiopubic ramus is present for examination, an accurate sex assessment using this feature alone can be achieved 77.8%. When dealing with burned remains, however, the morphology of the IPR may be reflective of sex, but the size of the IPR will almost certainly be reduced in size, rendering unusually small measurements that more likely fall within the female range. Using this method for badly burned pelves, therefore, is not recommended.

The shape of the IPR is a feature and morphology that is visualized during examination of the pelvis, is reflective of and part of the size and shape of the obturator foramen, and formation of the subpubic concavity. This part of the IPR is represented variously by a parallel, biconcave, or convex IPR and its narrowest width is often along the inferior ramus of the pubic bone but sometimes occurs at or near the junction (synchondrosis) of the ischiopubic bones. This study revealed that the narrowest width of the IPR varies from the pubic bone near the anterior portion of the obturator foramen to more inferior-posteriorly near the intersection of the two ischium and pubic bones. The authors found a few methods or mention of any use of the width of the IPR, visually or metrically, used to establish sex.15,16,25,31

While the width and depth of the IPR are incorporated into the size and shape morphology of the obturator foramen, this feature appears to have escaped mention as it relates to estimating sex of the pelvis. The morphology of the ischiopubic ramus is, however, an integral part

of the size and shape of the obturator foramen that is based, in part, on whether the IPR is convex, parallel, concave, or biconcave.

Several limitations should be noted. First, the interobserver technical error of measurement (9.6%) indicates moderate variability between observers, which may result from subtle differences in landmark placement and identification along the irregular contour of the ischiopubic ramus. Second, the validation set was relatively small (n = 24), which limits the generalizability of the present findings and warrants replication in larger and more diverse samples. Third, only left os coxae were analyzed; therefore, potential bilateral variation in the minimum ischiopubic ramus width remains untested. Future research incorporating both coxae and larger population samples would strengthen the reliability and applicability of this method.

Although machine learning models, including XGB, SVM, RF, ANN, and DT, were evaluated alongside traditional classifiers such as LDA and LR, the relatively small sample size of this study likely limited the ability of ML models to capture more complex patterns. The comparable accuracy observed among these methods indicates that classical statistical approaches remain reliable and efficient for sex estimation in small forensic datasets. Nevertheless, the inclusion of machine learning algorithms provides a comparative baseline and highlights their potential for future research using larger and more heterogeneous samples, where their capacity for non-linear modeling and feature interaction analysis could offer additional benefits.

CONCLUSIONS

This measurement provides an additional, straightforward tool that can be applied in addition to established sex estimation methods—such as the Phenice method, cranial and postcranial indicators—and may be particularly useful in cases involving fragmented or incomplete skeletal remains.

Data Availability Statement

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

ACKNOWLEDGEMENTS

The authors sincerely thank the curators, faculty, staff, and assistants for granting access to the collections and for their dedication to acquiring and preserving the materials referenced in this study. We also extend our appreciation to the two anonymous reviewers for their valuable comments and suggestions.

DECLARATIONS

Grants and Funding Information

None.

Conflict of Interest

The authors declare no conflicts of interest.

Registration number of clinical trial

None.

Author Contributions

Conceptualization and methodology, R.W.M.; Investigation, R.W.M., S.R., P.R., N.S.; Formal analysis, S.R.; Visualization and writing – original draft, R.W.M.; Writing – review and editing, R.W.M., S.R., P.R., N.S.; Supervision, R.W.M. All authors have read and agreed to the final version of the manuscript.

Use of Artificial Intelligence

During the preparation of this manuscript, the authors used ChatGPT to improve readability and grammar. All content was subsequently reviewed and verified by the authors, who take full responsibility for the final version.

REFERENCES

1. Klales AR. Sex estimation using pelvis morphology. In: Klales AR, ed. Sex estimation of the human skeleton: history, methods, and emerging techniques, Cambridge: Academic Press, 2020.

   p. 75-93.

2. Buikstra JE, Ubelaker DH. Standards for Data Collection from Human Skeletal Remains, Fayetteville: 1994.

3. Krishan K, Chatterjee PM, Kanchan T, Kaur S, Baryah N, Singh RK. A review of sex estimation techniques during examination of skeletal remains in forensic anthropology casework. Forensic Sci Int. 2016;261:165.e1-.e8.

4. Black V. Sex estimation using geometric morphometrics: evaluation of elements of the Pubis. Forensic Anthropology. 2021;4(3):47.

5. Waldron T. A report on the human bone from Merton Priory: English Heritage; 1985.

6. Mann RW, Koel-Abt K, Dhody A, Mahakkanukrauh P, Mann VJ, Techataweewan N, et al. The importance of human osteological collections: Our past, present, and future. Forensic Sci Int. 2021;325:110895.

7. Chomean S, Chatthai N, Sangchay N, Kaset C. Enhancing forensic sex identification through AI-based analysis of the foramen magnum. Forensic Science International: Reports. 2025;11:100411.

8. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2025. Available from: https://www.R-project.org/

9. Harris SM, Case DT. Sexual dimorphism in the tarsal bones: implications for sex determination. J Forensic Sci. 2012;57(2):295-305.

10. Knapp TR. Technical error of measurement: a methodological critique. American Journal of Physical Anthropology. 1992;87(2): 235-6.

11. Ward R, Jamison P. Measurement precision and reliability in

    
    

    craniofacial anthropometry: implications and suggestions for clinical applications. J Craniofac Genet Dev Biol. 1991;11(3):156-64.

12. Lovell NC. Test of Phenice’s technique for determining sex from the os pubis. Am J Phys Anthropol. 1989;79(1):117-20.

13. Suchey JM, Sutherland LD. Use of the ventral arc in pubic sex determination. J Forensic Sci. 1991;36(2):501-11.

14. Ubelaker DH, Volk CG. A test of the phenice method for the estimation of sex. J Forensic Sci. 2002;47(1):19-24.

15. Luo Y-C. Sex determination from the pubis by discriminant function analysis. Forensic Sci Int. 1995;74(1):89-98.

16. Dixit S, Kakar S, Agarwal S, Choudhry R. Sexing of human hip bones of Indian origin by discriminant function analysis. J Forensic Leg Med. 2007;14(7):429-35.

17. Blake KA, Hartnett-McCann K. Metric assessment of the pubic bone using known and novel data points for sex estimation. J Forensic Sci. 2018;63(5):1472-8.

18. Patriquin M, Steyn M, Loth S. Metric analysis of sex differences in South African black and white pelves. Forensic Sci Int. 2005;147(2-3):119-27.

19. Bytheway JA, Ross AH. A geometric morphometric approach to sex determination of the human adult os coxa. J Forensic Sci. 2010;55(4):859-64.

20. Gonzalez PN, Bernal V, Perez SI. Geometric morphometric approach to sex estimation of human pelvis. Forensic Sci Int. 2009;189(1-3):68-74.

21. Torimitsu S, Makino Y, Saitoh H, Sakuma A, Ishii N, Yajima D, et al. Morphometric analysis of sex differences in contemporary Japanese pelves using multidetector computed tomography. Forensic Sci Int. 2015;257:530.e1-e7.

22. Mahakkanukrauh P, Ruengdit S, Tun SM, Case DT, Sinthubua

    A. Osteometric sex estimation from the os coxa in a Thai population. Forensic Sci Int. 2017;271:127.e1-e7.

23. Vacca E, Di Vella G. Metric characterization of the human coxal bone on a recent Italian sample and multivariate discriminant analysis to determine sex. Forensic Sci Int. 2012;222(1–3):401. e1-e9.

24. Novotný V. Sex determination of the pelvic bone: a system approach. Anthropologie Paris. 1986;24(2-3):197-206.

25. Steyn M, İşcan M. Metric sex determination from the pelvis in modern Greeks. Forensic Sci Int. 2008;179(1):86.e1-e6.

26. Franklin D, Cardini A, Flavel A, Marks MK. Morphometric analysis of pelvic sexual dimorphism in a contemporary Western Australian population. Int J Legal Med. 2014;128(5):861-72.

27. Arun M, Nagesh K, Kumar GP. Estimation of sex from fragments of os coxa by metric analysis. Australian Journal of Forensic Sciences. 2012;44(2):145-53.

28. Sangchay N, Dzetkuličová V, Zuppello M, Chetsawang J. Consideration of accuracy and observational error analysis in pelvic sex assessment: A study in a Thai cadaveric human population. Siriraj Med J. 2022;74(5):330-339.

29. Lüthje P, Nurmi N, Kataja M, Heliövaara M, Santavirta S. Incidence of pelvic fractures in Finland in 1988. Acta Orthop Scand. 1995;66(3):245-8.

30. Ragnarsson B, Jacobsson B. Epidemiology of pelvic fractures in a Swedish county. Acta Orthop Scand. 1992;63(3):297-300.

31. Kane WJ. Fracture of the pelvis. In: Rockwood CA, Green DP, eds. Fractures in Adults, Philadelphia: J.B. Lippincott, 1984.p.1093-209.

31. Rogers T, Saunders S. Accuracy of sex determination using morphological traits of the human pelvis. J Forensic Sci. 1994; 39(4):1047-56.

‌Calcaneal Articular Talar Facets, Stieda’s Process, and Calcaneus Secundarius: Variations Found in Thai Population

Areeya Marie B. Wongla, B.Sc.1, Patara Rattanachet, Ph.D.1, Saranya Honghimaphan, B.Sc.1, Natipong Chatthai, M.Sc.1,2, Parawee Jitrabeab, M.Sc.1, Napakorn Sangchay, M.D., Ph.D.1,*

1Siriraj Anatomical and Anthropological Bone Research Centre (Virapan Davivongs), Department of Anatomy, Faculty of Medicine Siriraj Hospital,

Mahidol University, Bangkok, Thailand, 2Colleges of Medical Science, Western University Bangkok Campus, Bangkok, Thailand.

*Corresponding author: Napakorn Sangchay E-mail: napakorn.sac@mahidol.ac.th

Received 3 September 2025 Revised 4 November 2025 Accepted 9 November 2025 ORCID ID:http://orcid.org/0000-0001-9250-3702 https://doi.org/10.33192/smj.v77i12.277342

All material is licensed under terms of the Creative Commons Attribution 4.0 International (CC-BY-NC-ND 4.0) license unless otherwise stated.

INTRODUCTION

The calcaneus is irregularly cuboidal and the largest among the tarsals. It is located inferiorly to the talus, and therefore susceptible to fractures from high instant loads of force.1,2 Three articular facets of the calcaneus are described in anatomical nomenclature as: anterior, medial, and posterior articular facets.3 As part of the subtalar joint, these three facets aid in the maintenance of the foot’s longitudinal arch, alongside the talus, to prevent flatfoot abnormality.4-6 The talus is shaped similar to a saddle and is second to the largest of the tarsals. It articulates with the calcaneus, essential to joint movement.7 Just as the calcaneus noticeably has variations of its articular surfaces to the talus, so does the talus to the calcaneus— but with only two facets, namely the medial and posterior.8

Population-specific and cultural behaviors influence the occurrence of different facet types of the calcaneus and talus, such as squatting in Asia and Africa.9,10 Although the revelation is not new, the different shapes of facets could cause interference when studying parameters as the measurements of each calcaneal talar type do not have a comprehensive outline.11,12 Yang et al. (2019)12 quotes a study by Lawrence et al. (1989)13 that details how facet classification will be advantageous, especially in the fields of surgery and diagnosis. Nonetheless, there are studies that focus on the morphological features of the calcaneal facets that can link to sexual dimorphism.14-16

Foot pathologies are studied to understand their association with facet diversity.16-18 Syndromes of the foot can be linked to accessory bones, which may later fuse to bones as osteophytes. Skeletal foot variation supports trauma interpretation and population affinity estimations in anthropology15,19, leading to anatomical and morphological study categories applicable to various research fields.

In the calcaneus, the articular facets are extensively studied to investigate incidence and functional implications.5,10,12,20,21 Standard anatomy textbooks, like Gray’s Anatomy and Netter Atlas of Human Anatomy, and morphometric research analyze the superior portion of the calcaneus into patterns. However, three variants are generally recognized: fusion of the anterior and middle facets, separation of the anterior and middle articular facets, and absence of the anterior articular facet, as per the talus and calcaneonavicular joint.3,21-24 A trait called ‘calcaneus secundarius’ refers to the notch in the anterior articular facet filled by an accessory bone. In foot radiography, the trait is evident through diagnosis as the medial border of the anterior articular facet is “chipped off,” presenting an ossicle. In dry bones, the porosity and shape of the notch are considered, with or without the presence of the accessory bone.25,26 In metric studies, such as Ouamthong et al.27 and Scott et al.28, the measurements of length, breadth, and height of calcanei were conducted to determine applicability in

determining sex or population reference in forensic cases. This study pertains to only measure the articular facets.

In the talus, the diversities of the posterior process are subjected to more frequent research as opposed to the calcaneus secundarius.26 An elongated ossification from bone spur formation of the posterior lateral tubercle, is known as ‘Stieda’s process’. Named by German anatomist Ludwig Stieda in 1869, several studies have been conducted to learn more of the process’s influence on foot pathology and prevalence in sex and population. Through radiography, Stieda’s process is defined as having a lateral tubercle at least 5 mm in length.29 Other variants of the posterior process describe the shapes of the tubercle, including sharp, round, flat, and hook.30,31 The length of the posterior lateral tubercle will be considered for Stedia’s process and other variants, not solely based on visual observation.

This study aims to examine the prevalence of three calcaneal talar facet types, Stieda’s process, and calcaneus secundarius in the Thai population. It also investigates the correlation between limb dominance and sex to understand calcaneal and talar sexual dimorphism.

MATERIALS AND METHODS

Materials

A total of 250 specimens were used for this study, consisting of 250 pairs of dry calcanei and 250 pairs dry tali, each for male and female. The age at death of the samples ranged from 16 to 95 years in males and 16 to 99 years in females. The mean age in males and females is 69 and 68, respectively. The samples were chosen to exclude those with structural damage that could hinder parameter measurement, such as osteoporosis, severe spur, and postmortem damage. An incomplete set of bones, where either side of the calcaneus or talus is missing, is also part of the exclusion criteria. The bones were obtained from the bone collection of the Siriraj Anatomical and Anthropological Bone Research Centre (Si-AABRC), Department of Anatomy, Faculty of Medicine Siriraj Hospital, Bangkok, Thailand.

Methodology

From the third edition of Human Osteology22, three patterns of calcaneal talar facets are considered and used in this study for classification:

1. Pattern I— anterior and middle articular facets are fused together without constriction as a single facet, butwitha separatedposteriorfacet, consisting of two subtypes; (Fig 1a-1b)

2. Pattern II— anterior, middle, and posterior articular facets are separated and consist of three subtypes, determined by the distance between the anterior and middle articular facets; (Fig 1(2))

3. Pattern III—anterior articular facet is absent or too minimal in size to be considered a facet, while the middle articular facet and posterior articular facet stand independently. (Fig 1(3))

   In Pattern I, fusion of the anterior and middle facets meets the criterion of no gap between the two landmarks, and can sometimes be seen with a smooth “ivory-like, shiny patch” that can often be associated as eburnation.22 There are two subtypes that can visibly categorize Pattern I, “full-shaped” (FS) to characterize the fusion as one union with a continuous surface, and “waist-shaped/hourglass” (WH) to characterize the fusion of both the anterior and middle articular facets but a slight boundary or faintly “raised rims” can be seen to indicate the formerly two facets and is directly named by its shape (Fig 2).3,22,32

   To expand on Pattern II, the distances between the anterior and middle articular facets are categorized into subtypes derived from identical assessments done by Agarwal et al.31, Boyan et al.9 and Koshy, Vettivel, & Selvaraj30: Pattern IIa has a distance less than 2 mm; Pattern IIb has a distance between 2-5 mm; and Pattern IIc has a distance more than 5 mm— the distance is measured from the most posterior point of the anterior facet to the most anterior point of the middle facet (Fig 3).

   In Pattern III, an anterior articular facet of the calcaneus would articulate at the head of the talus (subtalar joint),

   
   

   Fig 1. Illustration of the patterns of calcaneal articular talar facets (right side): (1a) Full-shaped (FS) subtype of Pattern I; (1b) Waist-shaped/ hourglass (WS) subtype of Pattern I; (2) Pattern II with three subtypes determined by distance between anterior and middle articular facets;

   (3) Pattern III as the absence of an anterior articular facet.

   
   

   Fig 2. Pattern I calcaneal talar facet subtypes of the right calcaneus. See (1a) and (1b) in Fig 1 for illustration.

   
   

   Fig 3. Pattern II calcaneal talar facet subtypes of the right calcaneus depicting the least distance between anterior and middle articular facets to the most distant (left to right). See (2) in Fig 1 for illustration.

   
   

   not its neck (talocalcaneal neck articulation). Therefore, the absence of an anterior articular facet (AbAAF) is the most prominent feature of Pattern III (Fig 4).

   Stieda’s process (SP) refers to the extended ossification of the posterior lateral tubercle of the talus.33 When evaluating whether the SP is present, the talus is placed on a flat surface then viewed superiorly (Fig 5). If the posterolateral tubercle is enlarged or prominent in length, it is then marked present. For samples marked with the presence of SP, the lowest tip of the talar trochlea to the most posterior tip of the lateral talus process is measured to determine the average length (Fig 7). Absence of SP is preceded by noting down the anatomical variants derived from the criteria by Yang et al.34 and Kalbouneh et al.33 Four types are observed as follows:

   1. Sharp: lateral tubercle is a small, pointed projection;

   2. Round: lateral tubercle has a blunt, rounded end but short in size to not be mistaken as SP, with a groove for the flexor hallucis longus tendon;

   3. Flat: lateral tubercle and medial tubercle are faintly seen with the groove for flexor hallucis longus tendon shallow in appearance;

   4. Hook: lateral tubercle is wide and has a “neck” or the corner of the tubercle is chipped and curved towards or away from the groove for flexor hallucis longus tendon.

Fig 4. Pattern III calcaneal talar facet type of the right calcaneus. See

(3) in Fig 1 for illustration.

Calcaneus secundarius (CS) is evaluated by the crescent-shaped notch at the medial border of the anterior articular facet alongside an accessory ossicle.32 In the case that the ossicle is not found, the criteria to note down the presence of CS are the smooth roundedness/ triangulation of the notch’s shape and porosity at the notch.26,32,35 (Fig 6).

Morphometric Measurement

To determine the trait by sex estimation, the calcaneus and talus are subjected to measurement using a vernier caliper (mm). In Pattern I, the anterior articular facet

Fig 5. Illustration and photographs of Stieda process (SP) and the patterns of posterior lateral tubercle of the talus (right side): (1) Sharp type; (2) Round type; (3) Flat type; (4) Hook type.

Fig 6. Illustration and photographs of the calcaneus secundarius (CS) of the left and right calcaneus. Red asterisk indicates presence of CS.

(AAF) and middle articular facet (MAF) are fused together, therefore the conjoined facets (AMAF) are then measured from the most anterior point to the most posterior point (AMAFL), and the most medial point to the most lateral point of the fused facets (AMAFW). In Pattern II, AAF is measured by the length of the anterior articular facet (AAFL) and the width of the anterior articular facet (AAFW), applicable to MAF by measuring the length (MAFL) and width (MAFW) from its anterior border to posterior border of the facet. Since Pattern III has an absent anterior articular facet (AbAAF), only MAFW and MAFL were measured. For samples marked with the presence of SP, the lowest tip of the talar trochlea to the most posterior tip of the lateral talus process is measured to determine the average length (Fig 7).

Statistical analyses

Data is collected and submitted into Jamovi Version

2.5 (2024)36, an open statistical software, to analyze

Fig 7. Parameters of calcaneal articular talar facet measurements and SP measurement.

descriptive values of the calcaneus and talus including mean and standard deviation (SD). The data of the calcaneus and talus is assessed for normal distribution by the Shapiro-Wilk normality test. The formula to calculate the degree of sexual dimorphism (DoSD) was originated from Ricklan & Tobias (1986), which was used in former bone studies, namely the skull, scapula and clavicle.37,38

Xmale - Xfemale

7.2% displayed an average distance between facets at 1.66 mm, with an equal distribution between sexes; 19.6% showed a mean distance of 3.96 mm, occurring more frequently in females than males; and 10.4% exhibited a mean distance of 5.97 mm, equally comprised in males and females. Pattern III was the least frequent, observed in only 4.4% of samples characterized by the absence of an anterior articular facet, with higher occurrence in males than in females. In the odds ratio column, males

( Xmale

) × 100=degree of sexual dimorphism (%)

are more likely to feature Patterns I and II than females. However, the frequency of types indicates no significance as p-value >0.05.

To determine sex differences, the data is subjected to independent t-test, but if assumption of distribution is violated, then the Mann–Whitney U test is implemented. Siding (left and right side) of the calcaneus and talus to sex is assessed by paired t-test for normal distribution and Wilcoxon Signed Rank for violation of normal distribution. Chi-square test is applied for frequency data to determine the significance of prevalence. The value p < 0.05 determines statistical significance in this study.

Logistic regression analysis

A logistic regression equation was formulated using Posit Cloud39, formerly known as R Studio (Version 2025.5.1.513.3). The equation is developed by generalized linear model function – glm(). This model is then evaluated for the log of odd ratio (logit) and odds ratio (OR), where a logit equation for each side of a parameter was developed. The ROC curve was created where the model that presented with the lowest AIC (Akaike Information Criterion) and highest Area Under the Curve (AUC) was selected, by using pROC package.40 For each calcaneal pattern, the best parameters developed an equation where accuracy, AIC, and AUC were input into tables to determine the parameters with the most efficacy.

In the case of small sample number, the logistic regression models were computed to prevent overfitting by applying R package logistf by Heinze & Ploner (2024)41 into Firth’s bias-reduction method.42

RESULTS

The prevalence of calcaneal talar facet types is presented in Table 2. Pattern I was the most prevalent, accounting for 58.4% of samples. Within this pattern, 42.8% of specimens exhibited a full-shaped (FS) facet while 15.6% demonstrated an hourglass or waist-shaped (WS) facet, both subtypes with similar representation between males and females. Pattern II represented the second most common type, observed in 37.2% of cases. Among these,

In Table 3, the order of prevalence is similar to Table 2. Pattern I exhibits the highest prevalence of calcaneal talar types, accounted in 62.4% of samples. FS is found in 47.6% of samples with the frequency of females higher than males. WS is identified at 14.8% of samples, the number higher in females than males. Pattern II is in 33.2% samples. The first subtype exhibits in 4.8% of samples, an average distance between articular facets at

1.52 mm with 3 found in males and 9 in females; 20% of the samples have the second subtype with an average distance of 4.20 mm, 30 in males and 20 in females; the third subtype represents 8.4% of samples with an average at 7.06 mm between articular facets, comprising 13 in males and 8 in females. Pattern III is deemed the least varied, with a prevalence of 4.4% in samples found with 6 in males and 5 in females. By gathering the mean values of male and female parameters, Pattern II has the highest degree of sexual dimorphism at 8.38%. The odds ratio and significance of type frequency imply similar results to Table 2.

Sexual dimorphism statistics are summarized in Table 4. With the left and right sides of the calcaneus, eight out of sixteen variables have significant differences with p-value <0.05, which are: AMAFL (L&R) and AMAFW (L) of Pattern I, AAFW (L), MAFL (L&R), and MAFW (L&R) of Pattern II, and MAFW (L) of Pattern III. Together with Table 2 and Table 3, the average of sexual dimorphism degrees of the calcaneus are: Pattern I with 2.32%, Pattern II with 7.41%, and Pattern III with 6.68%.

In Table 5, the siding to sex prevalence of posterolateral tubercles showcases the flat variant to be the highest in prevalence at 41.4% or 207 tali which is followed closely by the round variant at 38.6% or 193 tali. The hook variant, or neck-like variant, is the least prevalent of types at 2.4% (12 tali) with the sharp variant higher at 7.4% (37 tali). SP is identified to be present in 10.2% of the talus samples (51 tali), which is the third highest prevalence of traits identified. The highest degree of sexual dimorphism found is the sharp posterior process

TABLE 1. List of abbreviations and descriptions.

Anterior articular facet (AAF) The anterior articular facet situated on the lateral surface of the calcaneus,

Middle articular facet (MAF)

The middle articular facet situated on the lateral surface of the calcaneus,

positioned medially.

positioned medially.

Anterior articular facet length (AAFL)

The distance from the anterior border to the posterior border of the anterior

articular facet

Anterior and middle articular facet A grouped abbreviation for Pattern I, FS and WS. (AMAF)

Anterior articular facet width (AAFW) The distance from the medial border to the lateral border of the anterior

Middle articular facet length (MAFL)

The distance from the anterior border to the posterior border of the middle

articular facet

articular facet

Middle articular facet width (MAFW) The distance from the medial border to the lateral border of the middle articular

Anterior and middle articular facet

length (AMAFL)

The distance from the anterior border of the anterior articular facet to the

posterior border of the posterior articular facet

facet

Anterior and middle articular facet The distance from the medial border to the lateral border of the fused anterior width (AMAFW) and middle articular facet, also known as the “width of constriction” according

to Mcshane43

Waist-shaped facet (WS) Fusion of the AAF and MAF of the calcaneus into one surface, either

continuous or shows signs of individual articular facets. The gap between AAF and MAF is filled but is minimal in size, presenting the articular facet to be shaped like an hourglass.

Stieda’s process (SP) An elongated projection of the posterior process of the talus, specifically the

Calcaneus secundarius (CS)

An accessory bone (ossicle) located near the anterior process of the

calcaneus with the ossicle fitting into the notch of AAF

lateral tubercle

TABLE 2. Contingency Tables for Calcaneal Articular Talar Facet Types and Prevalence of the Left Calcaneus.

Abbreviations: FS = Full-shaped facet; WS = Waist-shaped facet; AbAAF = Absent anterior articular facet

TABLE 3. Contingency Tables for Calcaneal Articular Talar Facet Types and Prevalence of the Right Calcaneus.

Abbreviations: FS = Full-shaped facet; WS = Waist-shaped facet; AbAAF = Absent anterior articular facet

Male

Female

Mean

P

DoSD Cohen’s

TABLE 4. Descriptive statistics, Shapiro-Wilk normality test of each parameter for both sexes and both sides of the calcaneus.

n = Sample size (pair of calcaneus); * = Shapiro Wilk (Normality Test): violation of normality assumption proceeds with Wilcoxon Signed Rank test. Bold text indicates significant difference.

TABLE 5. Prevalence and degree of sexual dimorphism of the posterolateral tubercle types to sex.

Posterolateral

tubercle types

Left

Right

Total

Prevalence

(%)

DoSD (%)

(n = 250) (n = 250)

Significance of Odds

prevalence ratio

Sharp M 8 9 37 7.4 15.0 0.599 0.835

Round

M

F

50

52

43

48

193

38.6

7.53

0.389

0.389

0.854

0.854

F 10 10 0.599 0.835

Flat M 55 49 207 41.4 0.96 0.885 0.974

Hook

M

F

0

3

4

5

12

2.4

1.0

0.240

0.240

2.040

2.040

F 51 52 0.885 0.974

Stieda’s process M 12 20 51 10.2 4.11 0.057 0.563

(SP)* F 9 10 0.057 0.563

*The length of Stieda’s process in this study ranges from 5.11 - 6.92 mm

of the talus at 15% and the least to be the hook posterior process. The presence is seen to be higher in males than females. However, all the four types of posterolateral tubercle types and SP do not have significant statistical difference nor significance in prevalence as all p-value exceeds 0.05.

Table 6 portrays the bilateralism and siding impact of the posterolateral tubercle types to the left and right sides of the talus. All traits have no significance to siding since the p-value exceeds 0.05.

The logistic regression analysis in Table 7 is conducted by standard logistic regression, and indicates the lowest AIC value and highest AUC value to be the most effective parameter to identify sex of the calcaneus by articular facet. In Pattern I, AMAFL has the strongest discrimination value, favoring the left side (AUC = 0.796). MAFL of Pattern II has the highest AUC values, with the left dominating over the right (AUC = 0.759).

Given the small sample size (n = 11), all analyses of Pattern III is considered exploratory until future research strengthens evidence of morphological differences between sexes in a bigger dataset. Firth’s logistic regression is computed to reduce bias as Pattern III has an insufficient amount of data. In single logistic regression, MAFW.L demonstrated the strongest relationship with sex (AUC

= 0.792), indicating a possible trend where greater width may increase the likelihood of being male while other variables showcased weaker associations (AUC ≤ 0.667).

DISCUSSION

There are considerable variations of anterior and middle articular facet patterns in different cohorts of people. Previous studies have also yielded results indicating a higher proportion of female samples to be more diverse compared to male samples, although the disparity may not be significant.44 Pattern I resulted with the highest incidence found amongst the samples, which is consistent with numerous studies conducted within Asian populations.4,6,11,14,31,45 In contrast, other populations exhibit Pattern II by the highest incidence, such as a Bulgarian population5, Anatolian population9, Turkish population44, and South Africans of European descent.15 European female samples have Pattern I as a majority whereas Indian and African female sample groups have a higher ratio of Pattern II.44 Pattern III is the least prevalent type of calcaneal talar facet, in agreement with a study dating back to 197411,45 and McShane’s (2021) comprehensive literature review in demographic studies, where the siding evaluation shows an average prevalence of 4%.43

Ancestral differences of the articular facets of the

calcaneus have been acknowledged because of repeated impressions in different populations, but no consensus has been formally established.17,46 A plausible explanation of how the lengths of each parameter have significance could be from how the fusion to a continuous facet is formed into two types, FS and WS. FS and WS are

TABLE 6. Bilateralism and siding significance of the posterolateral tubercle types of the talus.

Bold text indicates significant difference.

TABLE 7. Logistic Regression Model of calcaneus parameters for sex estimation.