A groundbreaking initiative is bringing together top minds in science, healthcare, and technology to tackle one of the most pressing health crises in the United States: opioid use disorder (OUD). With over 81,000 overdose deaths annually and 5.7 million Americans currently living with OUD, new approaches to prevention are urgently needed.

The Challenge: Relapse Detection Is Too Late

One of the key reasons opioid overdoses remain so deadly is that relapse—when a person in recovery starts using opioids again—is often unpredictable and detected only after it happens. By that time, it may be too late for effective intervention.

Currently, no standard clinical tools exist to monitor relapse risk in real-time in everyday life. This is a major gap in the continuum of care, leaving patients, families, and clinicians without timely information to act on.

The Solution: A New Risk Prediction Engine

The Digital Medicine Society (DiMe), in partnership with major institutions including Duke University, ŌURA, Google Fitbit, UNC Chapel Hill, and Alcohol and Drug Services, is launching a new project to fill this gap. Their goal? Build a risk prediction engine that uses real-time sensor data from wearable devices to identify early signs of opioid relapse.

This project will collect physiological and behavioral data from wearables and smartphones—such as:

• Elevated heart rate

• Insomnia or disrupted sleep

• Physical inactivity

• Psychological stress

• Social isolation

• Symptoms of anxiety and depression

These signals will feed into AI models, trained to detect patterns that often precede a relapse. The aim is to predict and prevent opioid use before it happens, empowering patients and care teams with timely, actionable insights.

Real-World Pilot and Research

The effort is co-led by Duke University’s BIG IDEAs Lab and will run an initial five-month pilot with patient participants to test the data quality and model performance. Importantly, the project is rooted in ethical and equitable principles, considering not only the science but also the real-world barriers to care like access, stigma, and mental health challenges.

Why It Matters

This initiative represents more than just a technological innovation—it’s a shift in how we approach addiction care. Instead of reacting after harm is done, this tool could offer a proactive, personalized safety net using technology that many already wear on their wrists.

As Shyamal Patel of ŌURA puts it:

“This can empower individuals to take proactive steps in their recovery, potentially reducing the burden on our overstretched public health systems.”

Source:

Digital Medicine Society (DiMe), 2025.
Project page: DiMe Risk Prediction Engine for OUD