TLDR: Mayo Clinic researchers have developed an innovative AI system capable of accurately identifying surgical site infections (SSIs) from patient-submitted postoperative wound photographs. Published in the ‘Annals of Surgery’, this two-stage AI model, trained on over 20,000 images, boasts 94% accuracy in detecting incisions and an 81% AUC for infection identification. This breakthrough aims to revolutionize postoperative care by enabling faster detection, reducing diagnostic delays, and streamlining communication between patients and care teams, especially crucial for outpatient and virtual follow-ups.
ROCHESTER, Minn. – A groundbreaking artificial intelligence (AI) system developed by a team of researchers at Mayo Clinic is set to transform postoperative care by accurately detecting surgical site infections (SSIs) from patient-submitted wound photos. The findings of this significant study were recently published in the prestigious ‘Annals of Surgery’.
The newly developed AI-based pipeline is designed to automatically identify surgical incisions, assess image quality, and flag potential signs of infection in photographs submitted by patients through online portals. This innovation addresses a critical need for timely outpatient monitoring of surgical incisions, a process traditionally time-consuming for clinicians and prone to delays in care.
Dr. Cornelius Thiels, a hepatobiliary and pancreatic surgical oncologist at Mayo Clinic and co-senior author of the study, emphasized the motivation behind the research: ‘We were motivated by the increasing need for outpatient monitoring of surgical incisions in a timely manner. This process, currently done by clinicians, is time-consuming and can delay care. Our AI model can help triage these images automatically, improving early detection and streamlining communication between patients and their care teams.’
The AI system operates using a sophisticated two-stage model. Initially, it determines whether a submitted image contains a surgical incision. Once an incision is confirmed, the model then meticulously evaluates the wound for any signs of infection. The ‘Vision Transformer’ model, central to this system, demonstrated remarkable accuracy, achieving 94% in detecting incisions and an 81% area under the curve (AUC) in identifying infections.
The system’s robust training involved an extensive dataset of over 20,000 images collected from more than 6,000 patients across nine Mayo Clinic hospitals, ensuring its reliability and broad applicability.
Dr. Hala Muaddi, a hepatopancreatobiliary fellow at Mayo Clinic and the study’s first author, highlighted the broader implications: ‘This work lays the foundation for AI-assisted postoperative wound care, which can transform how postoperative patients are monitored. It’s especially relevant as outpatient operations and virtual follow-ups become more common.’
Also Read:
- AI Enhances Surgical Precision: Text and Image Fusion for Critical View of Safety
- Japanese Study Reveals Retrieval-Augmented AI Eliminates Hallucinations in Clinical LLMs, Bolstering Patient Privacy and Accuracy
Researchers are optimistic that this technology will lead to faster responses for patients, significantly reduce delays in diagnosing infections, and ultimately support better care for individuals recovering from surgery at home. With further validation, the AI tool is envisioned to serve as a frontline screening mechanism, alerting clinicians to concerning incisions and potentially even detecting subtle signs of infection before they become visually apparent. This proactive approach could enable earlier treatment, decrease patient morbidity, and reduce healthcare costs.
