Modelling the seated human body’s response to whole-body vibration (WBV) remains challenging due to the complex influence of anthropometry, posture, and vibration characteristics. Conventional biodynamic models often rely on fixed weight assumptions and predefined body material properties, limiting their adaptability to real world conditions. This study therefore developed a machine learning (ML) based biodynamic model to predict vibration transmission across key body segments in seated shuttle car operators using field-measured vibration data. Six supervised ML algorithms were evaluated to identify the most suitable approach for predicting body segmental vibration transmissibility. WBV acceleration data were collected from 108 adult male shuttle car operators using triaxial accelerometers positioned at the seat interface and six body locations to capture vertical vibration transmission. Input predictors included age, body mass index (BMI), driving experience, posture type, seat-buttock interface clothing layers, excitation frequency, and vibration magnitude. The output targets were vibration transmissibility values at body segments. Model performance was assessed using mean absolute error (MAE), mean squared error (MSE), and coefficient of determination (R²) during training and testing, with additional validation using unseen datasets. Among all models, the artificial neural network (ANN) achieved the highest predictive accuracy (R² = 0.9839), followed by Gaussian Process Regression (R² = 0.9771), Decision Tree (R² = 0.9676), Random Forest (R² = 0.9582), and XGBoost (R² = 0.9569). Support Vector Regression showed comparatively poor performance (R² = 0.3121). The proposed ANN based biodynamic model reliably predicts vertical vibration transmission across seated human body segments and demonstrates strong potential for application in ergonomic assessment and WBV related occupational risk evaluation.
Shaikh et al. (2026) studied this question.