Background: Substance use disorders (SUDs), particularly opioid use disorder (OUD), remain a leading cause of mortality in the United States, with West Virginia bearing one of the highest overdose death rates nationally. Continuous monitoring of physiological precursors to relapse — including autonomic dysregulation, drug cravings, anxiety, and chronic pain exacerbation — represents a high-value but underutilized intervention target.

Objective: This study reports outcomes from a 100-participant prospective pilot of the CravAlert® program — an integrated wearable biosensor, cloud-based deep auto-encoder machine learning, remote patient monitoring (RPM), and just-in-time Peer Recovery Support Specialist (PRSS) intervention system that generates a unique digital phenotype for each patient, deployed across West Virginia Region 2.

Methods: One hundred individuals in early SUD recovery were enrolled across five community recovery and clinical sites in WV Region 2 (mean age 40; 63% female). Participants were monitored continuously via the VivaLink® VV330 FDA-cleared wearable biopatch, which streamed physiological data (HR, HRV, RR, skin temperature, GPS) to the VeeOne Health® deep auto-encoder machine learning analytics platform trained on over 1.2 million tagged data points. Condition-specific alerts triggered real-time PRSS interventions. Alert distributions, clinical outcomes at 3 and 6 months, and user experience survey results were analyzed.

Results: A total of 616 alerts were generated across the cohort. Anxiety and stress accounted for 398 alerts (64.6%), drug cravings 94 (15.3%), major depression 44 (7.1%), false positives 40 (6.5%), chronic pain exacerbation 22 (3.6%), relapses 14 (2.3%), and sleep apnea 4 (0.6%). At 3 months, CravAlert® participants demonstrated a relapse rate of 2% vs. 60% national average, and MAT retention of 74.5% vs. 46.2% national average. At 6 months, relapse rate remained 5% vs. 85% nationally, and MAT retention was 69.2% vs. 29% nationally. User experience surveys revealed 91% peer recommendation and 79% reported sense of safety.