Noncontact physiological evaluation of cognitive load among prolonged sitting workers using millimeter wave sensing

Heavy equipment operators on construction sites are required to remain seated in the cockpit for prolonged periods while performing work with high psychological intensity. Cognitive abilities are essential for these prolonged sitting workers to identify potential risk factors in the construction environment, as the accumulation of cognitive load may contribute to unsafe behavior. Traditional cognitive load monitoring methods rely on subjective surveys or invasive wearable sensors, which have limited applicability in construction sites. This study proposed a noncontact cognitive load monitoring method based on millimeter wave (mmw) radar. The cockpit environment of heavy equipment was simulated, and a 2-back task was designed based on the characteristics of the construction activities, where 20 participants were recruited and their physiological signals were continuously collected using mmw radar. Ten heart rate variability (HRV) features [e.g., root mean square of successive differences (RMSSD)] and six respiration rate (RR) features (e.g., mean) were extracted from the mmw signals to assess the workers' cognitive load. The features were learned and predicted with four intelligent classification algorithms. The results indicated that the extreme gradient boosting (XGBoost) algorithm performed the best, with an average accuracy of 93.8%. The proposed mmw sensing can monitor and assess the cognitive load of prolonged sitting workers in a noncontact manner, offering a potential tool for proactively managing and mitigating unsafe behaviors on construction sites.