Mayo Clinic Researchers Develop AI Tool to Detect Surgical Site Infections From Patient-Submitted Photos
Summary: Researchers at Mayo Clinic have developed an artificial intelligence (AI) system that analyzes patient-submitted wound photos to detect surgical site infections (SSIs) with high accuracy. The tool uses a two-stage pipeline to first identify incisions in the image, then flag signs of infection. It was trained on a dataset of over 20,000 images from more than 6,000 patients across nine Mayo hospitals. The AI achieved ~94% accuracy in incision detection and 0.81 area under the curve (AUC) for SSI detection. (Mayo Clinic News Network) :contentReference[oaicite:0]{index=0}
Key Highlights:
- Two-stage model: The pipeline first confirms presence of a surgical incision, then assesses infection risk within the incision area.
- Model performance: Incision detection accuracy ~94%; SSI detection AUC ~0.81 across validation sets. :contentReference[oaicite:1]{index=1}
- Diverse dataset: Trained across multiple centers, the system demonstrated consistent performance across racial groups, helping mitigate bias concerns. :contentReference[oaicite:2]{index=2}
- Clinical value: Automating image triage may reduce clinician workload, accelerate post-operative surveillance, and prompt timely intervention. :contentReference[oaicite:3]{index=3}
- Next steps: Prospective validation and integration into care workflows are underway. :contentReference[oaicite:4]{index=4}
Read the original article on Arab News
Keywords:
Hala Muaddi,
Cornelius Thiels,
Hojjat Salehinejad,
AI for SSIs,
postoperative monitoring,
wound imaging