University of Wisconsin Develops AI Tool to Identify Hospital Patients at Risk of Opioid Use Disorder

The AI system analyzes electronic health records in real-time, using pattern recognition techniques similar to how the human brain processes visual data.

The University of Wisconsin (UW) School of Medicine and Public Health has developed an artificial intelligence (AI)-driven screening tool designed to identify hospitalized adults at risk of opioid use disorder (OUD).

The tool is embedded within hospital workflows and aims to support timely addiction medicine consultations and withdrawal monitoring.

According to the university, the AI tool performs on par with traditional provider-led approaches in initiating consultations.

More importantly, patients flagged by the AI tool and referred for addiction medicine support were 47% less likely to be readmitted within 30 days of discharge, indicating substantial improvements in patient outcomes and cost savings.

The outcomes of the trial, supported by the National Institutes of Health (NIH), were recently published and demonstrate the practical benefits of integrating AI into clinical workflows.

Dr. Majid Afshar, principal investigator of the study and associate professor of medicine at UW, described the findings as a breakthrough in real-world applications of AI for addiction care.

"Our study represents one of the first demonstrations of an AI screening tool embedded into addiction medicine and hospital workflows, highlighting the pragmatism and real-world promise of this approach," Dr. Afshar stated.

He emphasized that AI could be a promising strategy for expanding access to addiction treatment, improving efficiency, and lowering healthcare costs.

How the AI Tool Works

The AI system analyzes electronic health records in real-time, using pattern recognition techniques similar to how the human brain processes visual data.

Once at-risk individuals are detected, the system automatically notifies healthcare providers with tailored recommendations for addiction consultations and withdrawal management.

The study reviewed 51,760 hospitalizations between 2021 and 2023. Of these, two-thirds occurred prior to the tool's deployment. A total of 727 addiction medicine consultations were conducted.

The AI-enabled group had a consultation rate of 1.51%, slightly higher than the 1.35% in the provider-only group. Notably, 30-day readmission rates were significantly lower in the AI group—8% compared to 14% in the traditional approach.

The National Institute on Drug Abuse, part of the NIH, provided funding for the project.

Stay tuned for more such updates on Digital Health News.