AI Tool Detects Surgical Site Infections from Patient-Submitted Photos

Surgical site infections (SSIs) remain a significant challenge in postoperative care, particularly as outpatient operations and virtual follow-ups become more common. The current method of monitoring surgical incisions often relies on clinicians reviewing patient-submitted photos, a process that is time-consuming and can result in delayed diagnosis and treatment. This lag in identifying infections may contribute to increased morbidity, patient anxiety, and unnecessary healthcare costs. To address these challenges, researchers have developed an artificial intelligence (AI) system that analyzes patient-submitted wound images for signs of infection and prioritizes cases needing clinical attention.

The AI system developed by researchers from the Mayo Clinic (Rochester, MN, USA) is designed to improve postoperative monitoring by detecting SSIs through photos uploaded by patients. The system, known as an AI-based pipeline, uses a two-stage model to first determine whether a photo contains a surgical incision and then assess the incision for signs of infection. Developed using over 20,000 images from more than 6,000 patients across nine Mayo Clinic hospitals, the model leverages a Vision Transformer architecture to process visual data. The research was motivated by the increasing demand for timely and efficient outpatient wound care, and the tool aims to streamline clinician workflows and enhance patient outcomes.

In a study published in the Annals of Surgery, the AI model demonstrated 94% accuracy in identifying surgical incisions and achieved an 81% area under the curve (AUC) for detecting infections. These results suggest that the technology could function as a frontline screening tool, alerting clinicians to concerning images and enabling earlier interventions. Importantly, the model performed consistently across diverse patient groups, addressing concerns about algorithmic bias. Researchers envision this tool supporting faster diagnosis, reducing patient complications, and prioritizing attention in rural or resource-limited settings. Prospective studies are currently underway to evaluate its integration into routine surgical care and determine its impact on long-term patient outcomes.

"This work lays the foundation for AI-assisted postoperative wound care, which can transform how postoperative patients are monitored," said Hala Muaddi, M.D., Ph.D., a hepatopancreatobiliary fellow at Mayo Clinic and first author. "For patients, this could mean faster reassurance or earlier identification of a problem. For clinicians, it offers a way to prioritize attention to cases that need it most, especially in rural or resource-limited settings."

Related Links:
Mayo Clinic