Using Re-inforcement Learning to Automatically Adapt a Remote Therapy Intervention (RTI) for Reducing Adolescent Violence Involvement

University Of Michigan At Ann Arbor, Ann Arbor MI

Investigators

Linked publications & trials

Abstract

Homicide is a leading cause of death for adolescents (age:14-24), disproportionately impacting African-American populations. Urban EDs are a critical opportunity for violence prevention, with >600,000 adolescents/year seeking treatment for violence-related injuries. In our study of violently-injured youth in urban EDs, we found that within 2-years, 37% returned for a repeat violent injury, 59% experienced firearm violence, 38% were arrested, and 1% died. Despite this, strategies to decrease repeat violence after an ED visit have not been well studied. Given our work demonstrating that single session ED interventions are efficacious reducing violence in lower risk adolescents, the application of this therapy, expanded to address greater problem severity over multiple sessions and enhanced by including care management, represents a potentially efficacious approach for altering risk trajectories of higher-risk violently-injured adolescents. Our recent pilot of this approach (S-RTI) was well received, addressing problems identified in prior multisession interventions (e.g., transportation) with the addition of remote therapy (e.g., phone). While innovative/promising, this S-RTI approach is resource intensive and does not address heterogeneity in treatment responses. By contrast, adaptive strategies allow for “just-in-time” tailoring that provides a balance between too much and not enough intervention and enhances outcomes while reducing cost. Reinforcement learning is an artificial intelligence domain that allows computer systems to “learn” from the success of prior treatments and is a promising approach to constructing adaptive “just-in-time” interventions. For this study, we are testing two versions of an RTI, a standard RTI condition (S-RTI) comprised of a single ED session followed by 5 remote sessions, and an adaptive RTI version (AI-RTI) optimized by reinforcement learning to step up/down the intensity of treatment between three levels (i.e., remote therapy, electronic bot messaging, assessment only) based on patient response to daily survey assessments. The original study aims were: 1) To refine/adapt our RTI for delivery using two packages (S-RTI; AI-RTI); 2) To conduct a 3- arm RCT enrolling 750 violently-injured adolescents seeking ED care (age:14-24) to compare efficacy of S-RTI (n=250), AI-RTI (n=300), and control (n=200); and, 3) To evaluate adaptability of the AI-RTI RL algorithm by comparing the first 50% of enrollees to the second 50% on process variables (e.g., engagement). Primary outcomes (6-, 12-months) include aggression and victimization. Secondary outcomes include ED recidivism for violent injury, substance use, mental health symptoms, and criminal justice involvement. While the current study holds promise for addressing elevated rates of violence, as well as key health disparities, among socio- disadvantaged youth, the clinical trial has experienced challenges stemming from COVID. This request for supplemental administrative funds is focused on adding a clinical recruitment site (to the currently enrolling sites), as well as clinical and research staff to avoid reducing scientific scope and to enhance the project’s ability to achieve the original study aims/goals, preserving the potential for high public health impact reducing violence.