Development of a real-time posture feedback system with integrated human motion prediction for proactive intervention in work-related musculoskeletal disorders

With the advancement of human-centric smart manufacturing toward Industry 5.0, the assessment and mitigation of work-related musculoskeletal disorder (WMSD) risks have become a critical issue in industrial workplaces. Although real-time posture feedback based on wearable devices has been increasingly adopted for ergonomic intervention, existing approaches mainly rely on reactive feedback and seldom consider the real-time optimization of continuous poor-posture sequences that trigger interventions. This study proposes a Human Digital Twin (HDT)–driven real-time posture feedback framework and develops a corresponding multimodal feedback system integrating visual, auditory, and haptic channels. The system continuously monitors workers’ postures using wearable sensing, evaluates postural risk levels in real time, and delivers personalized feedback to guide posture correction. Experimental validation demonstrates that the proposed system provides timely and effective ergonomic interventions, leading to a significant reduction in postural risk compared with conventional offline guidance methods. Furthermore, to address the limitation of reactive feedback strategies, a spatio-temporal Transformer-based human motion prediction algorithm (S2Transformer) is introduced to anticipate future postural evolution. The proposed algorithm predicts both skeletal joint positions and postural risk scores over short-term horizons, enabling proactive feedback generation before high-risk postures fully develop. Comparative experiments show that the proposed method achieves superior prediction accuracy compared with baseline models. This research establishes an HDT-driven proactive posture feedback paradigm by integrating multimodal real-time feedback with predictive human motion modeling, providing an effective solution for reducing WMSD risks in human-centric manufacturing environments.