Characterizing initial recovery from alcohol use disorder and predicting heavy drinking using mobile biosensors

PROJECT SUMMARY/ABSTRACT The overarching goal of this research project is to enhance the understanding and prediction of heavy drinking episodes during initial recovery (IR) from alcohol use disorder (AUD). Leveraging the latest advancements in wearable biosensor technology and artificial intelligence (AI), this project aims to address the significant gap in

current scientific knowledge regarding the fluctuating nature of AUD recovery. Our objective is to utilize cutting- edge wearable biosensors, like the VitalConnect VitalPatch and SCRAM-CAM, combined with sophisticated AI methodologies, including convolution neural networks and Hidden Markov Models, to elucidate the complex

interplay of physiological and neuroclinical factors in AUD recovery. Preliminary studies utilizing AI to analyze wearable sensor data have shown promising results in identifying behavioral phenotypes indicative of alcohol use risk, highlighting heart rate variability (HRV) as a particularly sensitive biosignal. The proposed research

encompasses two primary aims: 1) To develop real-time predictive models using wearable sensor data that can accurately forecast heavy drinking episodes during IR. By integrating data from electrocardiograms and transdermal alcohol sensors with traditional self-report and behavioral assessments, we aim to construct

comprehensive profiles that can anticipate periods of heightened drinking risk. 2) To identify and understand the neuroclinical and physiological mechanisms that contribute to or mitigate against shifts toward heavy drinking during IR. Through repeated biopsychosocial assessments and application of explainable AI

techniques, we intend to uncover critical factors influencing recovery trajectories. Crucially, this project is supported by a dynamic and interdisciplinary team, bringing together experts in AUD treatment research, wearable biosensor technology, and cutting-edge AI and computational methods. This collaborative approach

ensures a multifaceted perspective on the challenges of AUD recovery and enhances the project's capacity for innovative solutions and advanced data analysis. This project represents a novel integration of mobile health technologies and AI analytics in the study of AUD recovery. It has the potential to transform our understanding

of IR, leading to the development of personalized, just-in-time interventions for individuals battling AUD. The collaboration with a state-funded community-based outpatient clinic specializing in SUD treatment provides a unique opportunity to validate our models in a real-world setting, enhancing the potential for these findings to

be translated into practical, scalable monitoring and intervention tools.