Sex estimation from the greater sciatic notch: a comparison of classical statistical models and machine learning...

Area of Science:

• Forensic Anthropology
• Human Osteology
• Biometrics

Background:

• The greater sciatic notch (GSN) is a robust anatomical feature for sex estimation in skeletal remains.
• Its resistance to damage makes it valuable even in fragmented or degraded archaeological samples.

Purpose of the Study:

• To develop and compare statistical models for sex estimation using visual and metric analyses of the GSN.
• To evaluate the impact of incorporating additional variables on the accuracy of sex estimation models.

Main Methods:

• Analysis of 15 metric and 1 morphologic variables from 60 left coxal bones (30 male, 30 female).
• Comparison of six statistical models: Linear Discriminant Analysis (LDA), Regularized Discriminant Analysis (RDA), Penalized Logistic Regression (PLR), Flexible Discriminant Analysis (FDA), Support Vector Machine (SVM), and Artificial Neural Network (ANN).
• Models were tested using GSN variables alone and in combination with additional variables.

Main Results:

• High overall accuracy (0.92–0.97) achieved using GSN variables.
• Incorporating additional variables linked to the GSN significantly increased model accuracy.
• Penalized Logistic Regression (PLR) demonstrated superior performance after parameter optimization.
• The position of the deepest point of the GSN and the associated A2 angle are critical for sexual dimorphism.

Conclusions:

• Statistical models, particularly PLR, effectively estimate sex from the greater sciatic notch.
• Utilizing additional metric variables enhances the accuracy of sex estimation.
• The deepest point of the GSN and related angles are key indicators of sexual dimorphism in coxal bones.