Mandibular metric-based sex estimation in Asian populations: a novel OPLS-LDA framework

Area of Science:

• Forensic Anthropology
• Bioarchaeology
• Human Osteology

Background:

• Accurate sex estimation is crucial in forensic anthropology and bioarchaeology.
• Traditional methods face challenges due to population-specific biases and confounding factors.
• Need for robust, objective sex estimation techniques applicable to diverse populations.

Purpose of the Study:

• To develop and validate a novel sex classification method for Asian populations using mandibular metrics.
• To integrate Orthogonal Partial Least Squares (OPLS) for dimensionality reduction and Linear Discriminant Analysis (LDA) for classification.
• To establish a reliable framework for sex estimation in forensic and bioarchaeological contexts.

Main Methods:

• Analysis of 16 key mandibular parameters from dry mandibles (late 19th century).
• Application of OPLS to isolate sex-specific variations and reduce confounding factors.
• Univariate and multivariate LDA with cross-validation for classification accuracy assessment.

Main Results:

• Cross-validated accuracies for univariate LDA ranged from 75.7% to 87.4%.
• Cross-validated accuracies for multivariate LDA ranged from 82.5% to 85.4%.
• Maximum mandibular body height was the most significant predictor, achieving 87.4% accuracy.

Conclusions:

• The study presents a reliable framework for sex estimation in pooled Asian populations using mandibular parameters.
• OPLS effectively mitigated confounding factors, enhancing the robustness of LDA-based sex classification.
• Findings support the development of standardized sexing protocols for forensic applications across Asian populations.