Fitted models for the paper "A Hierarchical Bayesian Modeling Approach to Predict Upper Extremities' Fatigue in a Dynamic Order-Picking Task”.

Predicting upper extremities' fatigue in dynamic tasks is crucial for human reliability analysis and system safety in warehouses and distribution systems. New nonintrusive technologies, such as wearable sensors, can facilitate worker safety monitoring. This study proposes a Bayesian mixed-effects modeling approach that incorporates random slopes, intercepts, and non-linear terms to enhance prediction accuracy for upper extremities' fatigue. Data were collected from a laboratory experiment involving 14 participants who performed order picking tasks over four days under varying load and pace conditions. Inertial Measurement Unit sensors attached to the wrist, trunk, and upper arm recorded accelerometer and gyroscope signals. The proposed hierarchical model utilized task conditions, individual characteristics, and signal features, with prior distributions of parameters continuously updated over time. The results demonstrate that incorporating random effects alongside fixed effects significantly improves the prediction accuracy. Updating prior distributions over time effectively reduced prediction errors for the subjective fatigue indicators, thus supporting proactive interventions. Our approach enables more precise and individualized fatigue prediction in dynamic tasks, contributing to improved system safety in smart warehouses.