Mental Health Support and Crisis Detection for Veterans Using Opioids: Evaluating Predictive Modeling and Peer Support Through a Mobile App

Abstract

Veterans with Post-Traumatic Stress Disorder (PTSD) face heightened risks of opioid misuse due to the interplay between chronic pain management and psychological distress. The combination of PTSD and opioid use disorder (OUD) creates a complex clinical challenge, contributing to increased rates of overdose, suicide, and healthcare burden. Traditional interventions often fall short due to barriers such as stigma and fragmented treatment approaches. Peer support programs have emerged as a promising intervention strategy, leveraging lived experiences to improve engagement and adherence to treatment. This study evaluates the integration of predictive modeling techniques and peer mentorship through the BattlePeer mobile application, designed to support veterans with PTSD. We apply machine learning algorithms and rule-mining techniques to detect early indicators of mental health crises among veterans who use opioids. Our results suggest that a hybrid approach combining machine learning-based crisis detection with structured peer support can optimize mental health outcomes for veterans with PTSD and OUD.