@article{article_1789340, title={Sex Prediction Based on Mandibular Morphometry Using Machine Learning Algorithms}, journal={Meandros Medical And Dental Journal}, volume={26}, pages={481–489}, year={2025}, DOI={10.69601/meandrosmdj.1789340}, url={https://izlik.org/JA69YS29ZS}, author={Öztürk, Hakan and İpek, Eda Duygu and Tuzcu, Göksel}, keywords={Machine learning, Sex prediction, Sexual dimorphism, Mandibular morphometrics, Forensic anthropology}, abstract={Objective: The aim of this study is to investigate the potential of mandibular symphysis and mental trigone morphometry in determining sex in the Turkish population and to evaluate the performance of machine learning algorithms in sex prediction. Methods: Computed tomography (CT) scans of 350 adult individuals (191 males, 159 females) were retrospectively analyzed. Eleven morphometric parameters were measured from midsagittal and three-dimensional reconstructions of the mandible. Sex differences were assessed using independent t-test or Mann–Whitney U test, with ROC curve analysis performed to determine discriminatory accuracy and optimal cut-off points. ML models (SVM, KNN, Naïve Bayes, Logistic Regression, Random Forest, and XGBoost) were implemented in R. The dataset was split into 70% training and 30% testing sets, with 5-fold cross-validation and grid search applied for model optimization. Performance metrics included accuracy, sensitivity, specificity, F1-score, and AUC. Results: Ten of the eleven morphometric variables differed significantly between sexes (p <0.001), with males exhibiting larger dimensions. Symphyseal height (MSH, AUC=0.727), thickness (MST, AUC=0.726), and cross-sectional area (MSA, AUC=0.725) showed the highest discriminatory power. ML algorithms achieved strong classification performance (AUC range=0.821–0.851). Logistic Regression (F1-score=0.817, AUC=0.845) and XGBoost (F1-score=0.803, AUC=0.851) outperformed other models, while SHAP analysis identified PM_TML and MSH as the most influential predictors. Conclusion: Mandibular morphometry, particularly symphysis parameters, provides reliable indicators for determining sexual dimorphism. Analyzing these parameters using Logistic Regression and XGBoost methods offers a robust and interpretable methodological framework for forensic sex determination in the Turkish population.}, number={4}