The role of artificial intelligence in helping providers manage pain and opioid use after surgery

Excess opioid prescribing among post-surgical patients plays a major contributory role in the ongoing opioid epidemic in the United States—an epidemic that has devastating economic and human life tolls. Artificial intelligence (AI) and machine learning (ML) can be useful clinically by helping providers personalize opioid prescriptions based on individual patient needs and risk factors, thereby closing the gap between prescribed and consumed amounts. In this article, we explore the rationale for implementing AI/ML tools for post-surgical opioid prescription and recent use cases that have accurately predicted post-discharge opioid consumption and allowed providers to identify patients with unique prescription needs. These use cases not only address the challenge of capturing accurate and clinically relevant post-discharge patient data but also demonstrate promising preliminary evidence in reducing the opioid prescription volume at an institutional level. We later propose recommendations for AI/ML to be broadly applied across the entirety of the opioid prescription life cycle, from improving the precision of prescription quantity, monitoring individual- and population-level patient outcomes, to enhancing care integration. Lastly, we discuss potential challenges for future applications and scaling opportunities, including existing infrastructural limitations, evolving regulatory requirements, and the necessity of relevant stakeholder involvement throughout the intervention development process. In addition to ensuring the accuracy of model predictions, AI/ML-generated, data-driven insights also need to be reinforced by behavioral change strategies to encourage providers to adhere to recommendations. In short, AI/ML technologies can promote responsible and personalized opioid prescribing practices among healthcare providers, while achieving adequate post-surgical pain control and curbing opioid overprescription.