Intelligent Ergonomic Risk Assessment through Hybrid RNN–LSTM Modeling for Carpal Tunnel Syndrome Prediction

Abstract

Human activity recognition (HAR) has emerged as a critical research topic in intelligent healthcare systems and ergonomic risk analysis mainly due to the development of wearable sensor-based technologies and computing intelligence. This paper proposes a deep learning architecture that incorporates Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM) models for a dynamic, time-based analysis of ergonomic motion data in order to diagnose Carpal Tunnel Syndrome (CTS) early. Within the suggested framework, the RNN module is suitable for capturing short-run temporal dependencies, and the LSTM module is appropriate for capturing long-run contextual correlation, thus improving the process of learning sequential patterns. The RNN-LSTM architecture is found to be better in accuracy, robustness, and computational efficiency in identifying abnormal wrist and hand postures associated with the development of CTS. The research design involves data creation, multimodal signal acquisition, preprocessing, feature extraction, and end-to-end supervised model training. The experimental results have shown that the hybrid model minimizes data noise and false-positive alerts and enhances the prediction of CTS risk at an early stage. Additionally, the architecture facilitates real-time data processing and integration of adaptive feedback by enabling wearables with the Internet of Things (IoT). The suggested system creates a scalable base to identify repetitive strain injuries and leads to the development of smart and preventive healthcare and ergonomic monitoring solutions.