Sex classification accuracy through machine learning algorithms - morphometric variables of human ear and nose

Area of Science:

• Forensic anthropology
• Biometrics
• Machine learning applications

Background:

• Accurate sex determination is crucial for personal identification in forensic and medico-legal contexts.
• Facial features, specifically ear and nose morphology, offer potential biometric markers for sex estimation.
• Traditional methods may have limitations, necessitating novel approaches.

Purpose of the Study:

• To develop and evaluate a machine learning model for accurate sex prediction using ear and nose measurements.
• To identify the most significant facial parameters contributing to sex determination.
• To assess the efficacy of the PyCaret library in forensic sex classification.

Main Methods:

• A dataset of 508 participants (North India, aged 18-35) with recorded ear and nose measurements was utilized.
• The PyCaret machine learning library was employed, implementing a train-evaluate-test validation approach.
• Logistic regression was identified as the top-performing classifier after comparing multiple models based on accuracy and computational time.

Main Results:

• The logistic regression model achieved a sex prediction accuracy of 86.75%.
• Nasal breadth was identified as the most significant variable for accurate sex prediction.
• Most ear and nose measurements demonstrated significant contributions to sexual dimorphism.

Conclusions:

• Machine learning, specifically using PyCaret, offers a highly effective method for sex determination from craniofacial anthropometry.
• Nasal breadth is a key biometric indicator for sex estimation in forensic investigations.
• This approach can enhance personal identification in forensic examinations and crime scene investigations.