Predicting Sex from Hand Dimensions using Statistical Models: A Cross-Sectional Study of Medical Students

Determining sex from dismembered body parts is crucial for forensic and medico-legal investigations. Hand anthropometry offers a practical, non-invasive approach, particularly when utilising statistical models. This study aims to assess sexual dimorphism in hand dimensions and evaluate the effectiveness of statistical methods in determining sex from various hand measurements. Data for this study were obtained from a cross-sectional survey conducted from July to December 2021 involving a sample of 150 undergraduate medical students (78 males, 72 females) aged 18 to 24 years at a private medical college in India. Medical students were selected as a relatively homogeneous population to minimize confounding factors such as occupational variation, lifestyle, and health status, thereby increasing the internal validity of the findings. Measurements of hand length, breadth, and palm length for both the left and right hands were taken. Logistic regression was employed to develop models for sex classification. By using various combinations of explanatory variables, three logistic regression models were fitted to predict sex. Among these models, the one with the best fit was selected as the final model for sex prediction. It was observed that the mean scores of male and female respondents differ significantly. All hand dimensions were significantly larger in males (p < 0.001). According to the best-fitting model, the Right Hand Index (RHI) along with height was identified as the most significant predictors of sex. The best-fitted logistic model achieved 90% accuracy with an AUC of 0.93. Hand dimensions, particularly the Right Hand Indices (RHI), are effective predictors of sex and logistic regression provides a reliable method for forensic identification when complete body parts are unavailable, and this model can be utilised for other types of forensic predictions.