AI Tool Uses ECG to Predict Mortality Risk after Surgeries and Procedures

An artificial intelligence (AI) algorithm uses electrocardiograms (ECGs) to accurately predict how patients would fare after surgeries and procedures.

Researchers at the Smidt Heart Institute at Cedars-Sinai (Los Angeles, CA, USA) have trained the AI model to analyze pre-operative ECGs, uncovering a novel application for this test, which dates back to the late 19th century. An ECG, a standard test that records the heart's electrical activity by placing electrodes on the skin, helps assess heart function. The study included patients undergoing various surgical procedures, encompassing open heart surgery, major surgeries, and less invasive techniques using catheters or endoscopes.

The research team correlated pre-surgical or pre-procedural ECGs of the patients with their subsequent post-operative outcomes. They tasked the AI algorithm with detecting correlations or patterns within the ECG waveforms. While the algorithm classified most patients as low risk, it flagged others as high risk, revealing that these individuals had an almost nine times higher likelihood of post-operative mortality. Currently, physicians gauge a patient's surgery risk based on medical society guidelines. The investigators at Cedars-Sinai are exploring how to adapt this AI algorithm into a web-based application, aiming to make it broadly accessible to both medical professionals and patients.

“This is the first electrocardiogram-based AI algorithm that predicts post-operative mortality,” said David Ouyang, MD, a cardiologist in the Department of Cardiology in the Smidt Heart Institute at Cedars-Sinai. “Previously, algorithms have been used to assess long-term mortality as well as individual disease states, but determining post-surgical outcomes helps inform the actual decision to do surgery.”

“As it now stands, clinicians only have a modest ability to predict how a patient is going to do after surgery,” added Ouyang. “Current clinical risk prediction tools are insufficient. This AI model could potentially be used to determine exactly which patients should undergo an intervention and which patients might be too sick.”

Related Links:
Cedars-Sinai