Physiology-Informed Optimization of Work–Rest Schedules for Construction Workers Using Wrist-Worn Heart-Rate Monitoring

Abstract

In active construction settings, sustained physiological strain contributes to fatigue, injuries, and productivity losses, yet site schedules rarely adapt to workers’ real-time conditions. Most previous studies relied on fixed work–rest periods applied uniformly to all workers, and even when worker-specific scheduling methods were proposed, the physiological inputs were simulated or based on proxy datasets rather than measurements from active sites. Accordingly, this study develops a scheduling approach that uses wrist-based heart-rate reserve (%HRR) to trigger and size breaks, with a mixed-integer linear program (MILP) placing them under daily time and spacing limits, allowing break guidance to update continuously in response to on-site physiological signals. A live %HRR time series from three active sites is passed through a rest-allowance rule that identifies candidate breaks and estimates the recovery benefit of each. A MILP then selects a feasible subset of high-value breaks. In this evaluation, breaks were simulated rather than enforced on-site. The following week’s break schedule is then planned based on the results obtained using three different methods. Results show that 92% of worker-days needed at least one break, with 5−15 min/day being the most common total, and the average required rest ~40 min/day across workers. Under a 30−minute cap, the optimizer retained 84% of candidate breaks, typically by shortening rather than removing them, and reduced window-level mean %HRR by ~8 percentage points compared with no breaks. It also reduced long periods spent at high strain, (≥40% HRR). Field data showed distinct differences between workers and between sites, supporting the need for flexible, physiology-based scheduling. Thus, the approach provides supervisors with a practical way to plan rest windows using field data rather than relying on laboratory or simulation studies.