Identification of anthropometric variables influencing the prediction of manual strength using advanced models to prevent occupational injuries in the economically active population.

BackgroundHand grip and torque strength are critical in occupational tasks, particularly in industries requiring manual tool handling. Ergonomic designs tailored to workers' anthropometric characteristics can help reduce Musculoskeletal Disorders (MSDs), a leading cause of workplace absenteeism. While several studies have examined anthropometric influences on grip strength, predictive models integrating machine learning remain limited.ObjectiveThis study aims to identify the most influential anthropometric variables in predicting hand grip and torque strength among economically active individuals, using machine learning models to enhance force estimation for ergonomic applications.MethodsA cross-sectional study was conducted with 382 participants (194 women, 188 men), aged 15-65 years. Three predictive models-Linear Regression (LR), Random Forest (RF), and AdaBoost (AB)-were evaluated based on Mean Absolute Error (MAE), Mean Squared Error (MSE), and Explained Variance Score (EVS). Feature importance was analyzed using Mean Decrease in Impurity (MDI) for RF and AB and coefficient magnitude for LR.ResultsLR was the best model in four cases, while RF and AB performed better in nonlinear relationships. In women, grip and torque strength depended on hand and wrist dimensions, whereas in men, variables such as palmar length and lateral reach played a more significant role.ConclusionsAnthropometric characteristics significantly influence manual strength, highlighting the need for sex-specific ergonomic adaptations. Machine learning models improve predictive accuracy, supporting the design of interventions to reduce MSD risks.