An artificial intelligence tool developed with National Institutes of Health (NIH) funding has shown promise in identifying hospitalized adults at risk for opioid use disorder. This AI-driven method also suggests referrals to addiction specialists, providing a significant reduction in hospital readmissions compared to traditional methods. An NIH-backed clinical trial revealed that the AI tool was just as effective as healthcare providers in initiating consultations with addiction specialists and monitoring withdrawal symptoms. Patients who underwent AI screening had a 47% lower chance of readmission within 30 days post-discharge, with an estimated $109,000 in healthcare savings over the trial period.
Published in Nature Medicine, the study highlights AI's capability to improve patient outcomes in real-life healthcare settings, suggesting its potential as a cost-saving and efficiency-boosting investment for healthcare systems. Nora D. Volkow, M.D., director of NIH's National Institute on Drug Abuse (NIDA), noted the challenges of incorporating resource-intensive screenings in overwhelmed hospital settings but emphasized AI's potential to strengthen addiction treatment and optimize workflow while reducing costs.
The clinical trial, conducted by the University of Wisconsin School of Medicine and Public Health, assessed the performance of AI screening against provider-led initiatives between 2021 and 2023 at the University Hospital in Madison, Wisconsin. About 51,760 adult hospitalizations were screened, with the AI tool deployed in about a third of the cases. The AI screener analyzed electronic health records to identify signs of opioid use disorder and issued alerts to healthcare providers.
The trial showed that AI-prompted consultations were as effective as those initiated by providers, without losing quality, while offering a scalable solution. The study found that 1.51% of adults received consultations via AI tool assistance, compared to 1.35% without it. The AI tool also correlated with fewer 30-day readmissions, reducing such occurrences from 14% with provider-led methods to 8% with AI involvement.
After accounting for various patient demographics and conditions, the reduction in readmissions was significant. The researchers noted a potential decrease of 16 readmissions with the AI use, calculating a net cost saving of $6,801 per avoided readmission, totaling approximately $108,800 in savings during the study period.
While the AI screener demonstrated strong effectiveness, authors, including lead author Majid Afshar, M.D., of the University of Wisconsin-Madison, cautioned about alert fatigue among providers and called for broader validation in diverse healthcare systems. They also acknowledged the potential biases due to the evolving opioid crisis. Future research is needed to refine AI integration and evaluate its long-term impact on patient care outcomes.
The AI tool emerges as a promising intervention amid the ongoing opioid crisis, which saw a 6% rise in emergency department admissions between 2022 and 2023. Despite the effectiveness of AI in early intervention, continued research is necessary to optimize its application in healthcare environments.
Help is available for those struggling with opioid use disorders; resources include FindSupport.gov, FindTreatment.gov, and the 988 lifeline.
