Facial Expression-Based Monitoring of Fatigue and Stress in Care Facility Workers

This study explores the relationship between objective facial expression–based indicators and subjective perceptions of fatigue and stress among care workers in Japan. Caregiving involves significant physical and emotional demands, yet traditional survey-based assessments are often constrained by limited time and resources in care settings. To address this challenge, the study employed video recordings and automated facial analysis to obtain objective indicators such as the Percentage of Eyelid Closure over Time (PERCLOS) and selected Facial Action Units (AUs). These metrics were compared with structured questionnaire responses regarding perceived fatigue and stress. The results revealed consistent links between objective facial metrics and self-reported experiences. Elevated PERCLOS values were observed in participants who reported heavy workloads and task-related difficulties, suggesting that increased eye closure reflects physical fatigue. Higher scores for AU04 (brow lowerer) and AU07 (lid tightener)—markers of muscular tension—aligned with perceptions of sustained effort. A composite AU index comprising AU01, AU02, AU20, and AU23 corresponded with emotional exhaustion and anxiety. One notable case exhibited both the highest AU composite scores and elevated self-reported stress, supporting the potential of facial metrics as indicators of psychological strain. These findings demonstrate that AU intensity values, rather than binary presence indicators, enable the detection of subtle expressions linked to fatigue and stress. While no single metric proved superior, integrating facial metrics with subjective assessments offers a more holistic view of caregiver well-being. Overall, this study highlights facial expression analysis as a promising non-invasive tool for early detection of fatigue and stress, with practical implications for workplace interventions and future longitudinal research.