The broader impact/commercial potential of this I-Corps project revolves around the human brain?s proclivity towards substances detrimental to human health and wellbeing, such as dangerous amounts of sugar, salt, and drugs, which is causing global health and safety risks. Inability to avoid these temptations shortens lifespans, increases healthcare costs, and strains resources. This team will use machine learning and pattern recognition AI to identify and react to factors that result in destructive human behaviors. While the initial intent of our product is to provide potential customers with a tool to detect and prevent addiction overdose and relapse, offer interventions of supervised individuals, conserve emergency response resources, and save taxpayer money, this unsupervised learning AI could, in subsequent iterations, be used to identify and react to any behavior -- such as unsafe driving, binge eating, or anger management. Helping people identify the precursors to their behaviors could serve as a type of biofeedback that gives them and their support networks timely and individualized insights, preventions, and interventions -- some of which could be implemented by Internet of Things enabled devices (e.g. locking a car?s ignition; calling a sponsor; playing calm music; etc.), resulting in cost and health benefits across many populations.
This I-Corps project is focused on a patent-pending platform to predict and prevent addiction relapses and overdoses by treating those in recovery with timely interventions using wearable devices and artificial intelligence. Detecting craving states is supported by past and current research, and there is scientific justification for gathering smartphone usage and biometric wearable data on drug users. Researchers have used the combination of smartphone usage and wearables data to predict binge drinking behavior and we plan to apply this research to opioid use. Additionally, cocaine and opioid research uses the same wearable device as our studies. We expand upon current research by combining just-in-time smartphone-based interventions for people in recovery for opioid use disorders with pattern-detection predictive models that will be trained at the population level and refined at the individual level. We will gather data on all relevant factors that determine drug relapse risk state: physiology, smartphone usage, location data, self-reports, and support-network-reports. Once a high-risk state is detected, a smartphone app-based intervention focused on behavior change will then be implemented in a timely and customizable manner.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
