AI tool determines postoperative mortality risk

Editor's Note

A new artificial intelligence (AI) tool developed by researchers at the Smidt Heart Institute at Cedars-Sinai could help doctors better understand which patients are at greatest risk of dying following surgery, according to a December 15 report in Newswise.

Based on findings originally published in the Lancet Digital Health, report describes the tool’s significance as the first electrocardiogram-based AI algorithm that predicts post-operative mortality. As opposed to previous models used to assess long-term mortality and individual disease states, determining post-surgical outcomes helps inform the decision of whether to perform surgery, researchers say.

The AI algorithm drew on data for 45,969 Cedars-Sinai Medical Center patients who had a complete ECG image available before their procedure date between January 1, 2015 and December 31, 2019, and paired this information with patients’ preoperative ECG. AUC, or area under the curve—a measure of machine learning algorithm performance—was 0.83, indicating that the tool performed well in predicting postoperative mortality in various including cardiac, non-cardiac and catheterization-based procedures. The AI tool classified most patients as low-risk for mortality, but those identified as high risk had an almost 9-fold increased probability of postoperative mortality.