Wearable-derived measures of stress physiology in patients with opioid use disorder treated with methadone: A stress-task study in a real-world clinic with healthy controls.

内容摘要

Wearable biosensors enable real-time monitoring of physiological stress. While methadone maintenance treatment (MMT) is effective in opioid use disorder (OUD), inter-dose withdrawal and stress dysregulation remain challenges. We sought to identify wearable-derived digital biomarkers of stress in individuals with OUD receiving MMT. In this cross-sectional, laboratory-based study, 72 adults completed 76 sessions (14 OUD pre-methadone dose, 15 OUD post-dose, 47 healthy controls) of a standardized protocol of calm tasks (e.g., relaxing videos, positive memory recall) followed by stress-inducing tasks (e.g., horror video, mental arithmetic, negative memory recall). Physiological data were continuously recorded using the Empatica E4 wristband (heart rate, heart rate variability [HRV], electrodermal activity [EDA], skin temperature, accelerometry). Features were extracted using open-source toolkits. Linear mixed-effects models tested calm-stress differences and group effects; Random Forest classifiers evaluated calm versus stress classification. Among 340 extracted features, 63 (18.5%) significantly differentiated calm from stress across all groups, with EDA- and accelerometry-derived features showing the most consistent stress sensitivity. Compared with controls, OUD participants demonstrated higher baseline arousal and altered stress reactivity, with the largest responses observed in the pre-dose group. Random Forest models performed best for cognitively demanding tasks (up to 98% accuracy) and more modestly for other tasks. Wearable sensors can detect multidimensional alterations in resting-state physiology and stress reactivity among individuals with OUD on methadone. Future studies should validate these candidate biomarkers and test whether integrating wearable-derived stress detection into treatment improves clinically meaningful outcomes, such as stress management, craving, risk of return to opioid use, and retention.