AI Screening Tool Proves Effective in Identifying and Treating Opioid Use Disorder in Hospitals

AI Screening Tool Revolutionizes Opioid Use Disorder Care in Hospitals

A groundbreaking clinical trial, supported by the National Institutes of Health (NIH), has demonstrated the effectiveness of an artificial intelligence (AI)-driven screening tool in identifying and treating opioid use disorder (OUD) among hospitalized adults. The study, published in Nature Medicine, showcases how AI can match the performance of healthcare providers while improving patient outcomes and reducing healthcare costs 1

1

Study Design and Implementation

Researchers at the University of Wisconsin School of Medicine and Public Health conducted a clinical trial comparing physician-led addiction specialist consultations to an AI screening tool. The study, which screened 51,760 adult hospitalizations, was carried out in three phases:

1. Pre-intervention period (March-October 2021 and 2022): Provider-driven ad hoc consultations
2. AI tool development and refinement
3. Post-intervention period (March-October 2023): Hospital-wide deployment of the AI screener

The AI screener analyzed electronic health records in real-time, identifying patterns associated with OUD and issuing alerts to providers with recommendations for addiction medicine consultations and withdrawal management 2

2

Key Findings

The study revealed several significant outcomes:

1. Consultation rates: 1.4% of hospitalized adults received addiction medicine consultations with AI assistance, compared to 1.3% without it, demonstrating non-inferiority 1

   1

   .

2. Reduced readmissions: Patients screened by AI had 47% lower odds of 30-day hospital readmission compared to those with provider-initiated consultations 2

   2

   .

3. Cost savings: The reduction in readmissions translated to an estimated $108,800 in healthcare savings during the eight-month study period, even after accounting for AI software maintenance costs 2

   2

   .

Related Stories

Large-scale clinical trial finds generative AI improves care quality but not patient outcomes

AI Model Quadruples Delirium Detection in Hospitals, Improving Patient Outcomes

AI-Powered Alerts Significantly Improve Suicide Risk Detection in Neurology Clinics

AI Models Show No Bias in Opioid Treatment Recommendations, Study Finds

Implementation Process and Challenges

The researchers employed a hybrid effectiveness-implementation framework to optimize the AI screener's utilization:

1. Interviews with healthcare providers to identify potential barriers
2. Two rapid Plan-Do-Study-Act (PDSA) cycles to refine the system
3. Updates to the Best Practice Alert (BPA) to improve information provided to addiction specialists

While the AI screener was well-received by users, some providers expressed concerns about alert fatigue, particularly in high-demand settings 1

1

Implications and Future Directions

Dr. Nora D. Volkow, Director of NIH's National Institute on Drug Abuse (NIDA), emphasized the potential of AI to strengthen addiction treatment implementation while optimizing hospital workflow and reducing healthcare costs 2

2

Lead author Dr. Majid Afshar highlighted the study's significance as one of the first demonstrations of an AI screening tool embedded into addiction medicine and hospital workflows 2

2

Future research will focus on optimizing the AI tool's integration and assessing its longer-term impact on patient outcomes. The study underscores the potential of AI in addressing the ongoing opioid crisis and improving healthcare delivery in real-world settings.