An Automatically Adaptive Digital Health Intervention to Decrease Opioid-Related Risk While Conserving Counselor...

Area of Science:

• Artificial Intelligence in Healthcare
• Digital Health Interventions
• Reinforcement Learning Applications

Background:

• Opioid analgesics (OA) pose risks like dependence and overdose.
• Large-scale risk reduction for OA is challenging due to impractical counseling requirements.

Purpose of the Study:

• Evaluate a reinforcement learning (RL) AI intervention for personalized patient engagement.
• Assess if the AI can decrease self-reported opioid misuse behaviors.
• Determine if the AI can conserve counselor time.

Main Methods:

• A digital health intervention (PowerED) used RL to personalize weekly interactions for 228 patients with pain.
• RL algorithm selected from brief/extended motivational IVR calls or live counselor calls.
• The goal was to minimize OA risk and conserve counselor time by favoring IVR when effective.

Main Results:

• The RL intervention adapted over time, significantly reducing live counseling sessions.
• This adaptation led to progressively greater improvements in self-reported OA risk scores.
• Patients with the highest baseline risk showed particularly pronounced improvements.

Conclusions:

• Reinforcement learning effectively identified optimal treatment modalities for reducing opioid misuse behaviors.
• RL-supported digital health interventions offer a scalable solution for managing OA prescriptions.