Prediction of Sex and Body Mass Index Based on Anthropometric Features and Balance Parameters Using Machine Learning

Machine Learning-Based Prediction of Sex and BMI

Abstract

Background: This study aims to evaluate the predictability of individuals’ Body Mass Index (BMI) and sex by integrating anthropometric data and balance parameters into machine learning algorithms. The primary goal is to propose a faster, more reliable, and data-driven model compared to traditional statistical approaches. Additionally, the effects of BMI, height, and weight variables on balance performance were analyzed.

Materials and Methods: This study included 210 healthy adults (105 males, 105 females) aged 18–25 years. Height and weight were measured following standard anthropometric protocols, and balance performance was evaluated via the single-leg stance test on both sides. Anthropometric parameters (height, weight, dominant side) and balance durations were analyzed using machine learning algorithms (Decision Tree, Random Forest, Support Vector Machine). Model accuracy, precision, recall, and F1-score were calculated through 10-fold cross-validation. Statistical analyses were performed with IBM SPSS Statistics 25.

Results: Machine learning algorithms achieved high accuracy rates in predicting both BMI and sex. The Random Forest model demonstrated the best overall performance, reaching an accuracy of 84% for sex prediction and 98% for BMI prediction. Furthermore, our results revealed that BMI and body weight negatively affected balance performance, whereas height was not identified as a significant determinant.

Conclusion: The integration of anthropometric and balance parameters with machine learning algorithms provides an effective and non-invasive approach for predicting BMI and sex. This hybrid biomechanical–computational model may serve as a promising tool in personalized health assessments, balance analysis, and early risk detection studies.