AI and Medical Experts Unite to Predict Chronic Post-Surgical Pain

In a groundbreaking collaboration, researchers at Washington University in St. Louis have joined forces with AI scientists to tackle a significant medical challenge: persistent post-surgical pain. This condition affects approximately 10–35% of the estimated 300 million individuals who undergo surgery globally each year. The complex interplay of biological, psychological, and emotional factors makes it difficult to predict and address this widespread issue.

Persistent post-surgical pain is characterized by enduring discomfort that remains long after the surgical site has healed. According to Simon Haroutounian, a professor of anesthesiology at Washington University School of Medicine, the intricacies of this pain arise from a combination of surgical trauma and the interactions within the peripheral and central nervous systems, coupled with a person’s cognitive and emotional response to pain.

To enhance predictive capabilities, the research team, led by Haroutounian and Chenyang Lu, director of the AI for Health Institute, has developed machine learning algorithms that analyze preoperative data to identify patients at risk for chronic pain. Their findings were published in the Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies on June 20, 2025.

The machine learning model not only predicts the likelihood of persistent pain but also provides uncertainty estimates for these predictions. Ziqi Xu, a Ph.D. student in Lu's lab and the study's lead author, explained that this innovative approach allows the model to communicate its confidence level regarding risk assessments. For instance, a prediction might indicate that a patient has a 30% chance of developing persistent pain, with a 50% uncertainty in that estimate. This nuanced understanding assists physicians in making informed decisions regarding patient care.

The research involved 780 patients, who completed daily surveys via smartphone leading up to their surgeries. The data collected, combined with clinical histories and lab results, enabled the researchers to create a robust risk assessment model, which demonstrated better performance compared to existing prediction algorithms.

In their analysis, the researchers found that while some patients' pain may stem from behavioral factors, others might experience pain due to an immune response following surgery. This differentiation is crucial for tailoring appropriate interventions, which may range from cognitive behavioral therapy to strategies aimed at modulating immune responses.

As the team progresses with their research, the goal is to integrate the machine learning model into clinical practice, thereby enabling healthcare providers to predict and manage post-operative pain more effectively. They aim to facilitate a deeper understanding of the factors contributing to individual pain experiences, ultimately leading to personalized interventions that could significantly reduce the burden of persistent post-surgical pain.

This pioneering work exemplifies how collaboration between AI technology and medical expertise can transform patient care. The implications of these findings extend beyond individual patients, potentially reshaping surgical practices and pain management protocols on a global scale. As the research evolves, the focus will remain on enhancing predictive accuracy and developing targeted interventions that could alleviate suffering for millions of surgical patients worldwide.