RETRACE: AI Model Estimates Opioid Craving from Wearable Data
A team of researchers has introduced RETRACE, an AI framework designed to assess opioid cravings using physiological data from wearables, tackling the difficulties in aiding interventions for opioid use disorder (OUD). The findings, published on arXiv (ID: 2608.14947), indicate that stress triggers significant autonomic reactions, whereas craving signals are less pronounced and intertwined with stress responses. Psychological resilience, which influences both stress management and susceptibility to cravings, is represented through reusable indicators such as heart-rate recovery after stress. RETRACE leverages these indicators to improve craving assessment in subject-independent evaluations, enhancing detection in practical environments. The paper outlines empirical analyses and the model's structure, highlighting the fusion of psychological elements with physiological tracking, advancing digital health and AI-based addiction treatment for tailored care and remote observation.
Key facts
- RETRACE is a resilience-guided trait-conditioned model for craving estimation.
- The study is available on arXiv with ID 2608.14947.
- Stress elicits strong autonomic responses, while craving signals are weaker and embedded in stress physiology.
- Psychological resilience is captured via subject-level proxies like post-stress heart-rate recovery and autobiographical memory recall.
- The model addresses subject-independent evaluation challenges.
- The research targets opioid use disorder (OUD).
- Wearable physiological signals are used for detection.
- The work aims to support proactive interventions.
Entities
Institutions
- arXiv