Leiden’s PERISCOPE AI model predicts postoperative infection risks

Researchers at Leiden University Medical Center (LUMC) have developed PERISCOPE, an AI-driven tool designed to predict the risk of infections in postoperative patients. By analyzing nearly a decade of anonymized patient data, including vital signs and medical history, PERISCOPE assesses the likelihood of infections occurring within 37 days after surgery.

Tested on over 250,000 surgical cases across LUMC, Radboudumc, and Ziekenhuis Oost-Limburg, the model demonstrated performance comparable to experienced clinicians and outperformed less experienced staff, particularly in uncertain cases. The tool presents risk assessments as percentages and categories (low, medium, high) on a dashboard integrated with electronic patient records, streamlining decision-making and potentially reducing hospital stays and readmissions.

Currently undergoing integration into LUMC’s electronic health record system, PERISCOPE is expected to be operational by mid-2026. Future enhancements aim to enable preoperative risk predictions and specify infection types, further supporting clinical decisions and improving patient outcomes.

Read more here