EMG-FLES: Flexible wearable electromyography and edge-AI system for near-real-time muscle fatigue monitoring for construction-relevant tasks

Muscle fatigue is a major contributor to safety risks and productivity loss in construction. Although electromyography (EMG) provides a direct physiological marker of localized fatigue, continuous in-field monitoring remains challenging due to bulky instrumentation, unstable electrode–skin contact, and limited real-time deployable analytics. This study proposes EMG-FLES (Flexible Wearable Electromyography and Edge-AI System), a wearable sensing and mobile-edge inference framework for near-real-time muscle fatigue monitoring. EMG-FLES integrates a flexible skin-conformal EMG sensor, wearable wireless streaming hardware, and on-device TinyML inference to enable continuous, untethered muscle fatigue assessment. Evaluated on construction-relevant material-handling tasks targeting the biceps muscle group, the system achieved a mean fatigue classification accuracy of 85% with low per-window computational latency, enabling timely fatigue alerts during task execution. These results demonstrate the feasibility of combining flexible wearable biosensing with edge intelligence to support near-real-time fatigue monitoring in construction-relevant environments.