Real-time Detection of Psychological Fatigue in Middle-Aged and Elderly Individuals During Physical Exercise Using Wearable Devices

Abstract

As the global aging population continues to accelerate, exercise-related health management for older adults has emerged as one of the key research priorities in the field of public health. Psychological fatigue incurred during physical exercise is a critical factor contributing to the interruption of exercise activity among older adults, an increased risk of exercise-related injuries, and the development of exercise-related burnout; however, conventional subjective assessment methods and offline monitoring approaches are often inadequate for the continuous and real-time detection of psychological fatigue states. This study develops a real-time detection system for exercise-related psychological fatigue in middle-aged and elderly individuals, leveraging multimodal wearable devices and deep learning technologies. This system collects physiological signals-such as heart rate variability, three-axis acceleration, and skin conductance response-during physical exercise in elderly individuals. Following data preprocessing and feature extraction, it constructs a multimodal deep learning model combining temporal features, namely a CNN-LSTM architecture, to enable quantitative assessment and real-time prediction of psychological fatigue status. Experimental results demonstrate that this model achieves a prediction accuracy of 82.3% for psychological fatigue in middle-aged and elderly individuals, with a root mean square error (RMSE) of 1.25 and a Pearson correlation coefficient (r) of 0.81, significantly outperforming traditional single-feature linear regression models. This system provides an objective technological solution for intelligent and personalized monitoring of exercise-related health in middle-aged and elderly populations, while also offering data support for the development of exercise intervention strategies within the field of smart healthcare.