Publication: 3d sex prediction model in malays using lower second molar
Loading...
Date
2026-02
Authors
Hong, Tan Zong
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Sex prediction models play a crucial role in forensic science, particularly in disaster victim identification (DVI), by reducing the potential identification pool by approximately half. Sex is also a fundamental component of the biological profile, as it forms the basis for estimating other parameters such as age and stature. To date, no studies have investigated sex prediction using occlusal traits—specifically intercuspal distance (ICD), interpit distance (IPD), and total length of the central groove (TLCG)—within the Northeastern Malay population of Malaysia. Therefore, this study aimed to evaluate the relationship between occlusal traits of the lower second molar and sex among the northeastern Malay population. Measurements of ICD, IPD, and TLCG were obtained indirectly from 3D dental casts of 200 Malay individuals (100 males and 100 females) using 3-Matics software. An independent t-test was conducted to assess sexual dimorphism between sexes. Sex prediction models were developed using four machine learning approaches: Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), and Decision Tree (DT). The results demonstrated that mean values of ICD, IPD, and TLCG were generally higher in males than in females, except for the ML–MB distance. Among the models tested, LR achieved the best performance, with an area under the curve value (AUC) of 0.70. Stepwise LR analysis identified ML–MB (p = 0.008), MB–DB (p = 0.014), and MB–DL (p = 0.007) as significant predictors of sex. Two of the variables studied were statistically significantly larger in males than females. The percentages of sexual dimorphism ranged from 0.3-7.4%. The highest sex prediction model was LR and yielded fair classification (classification accuracy = 0.63, AUC value = 0.70). Other models yielded poor classification.
Description
Keywords
-