In the latest study, the researchers applied multiple AI-derived cardiac biomarkers generated by ECG Buddy to preoperative risk assessment.
Because anesthesia and surgery place additional stress on the cardiovascular system, accurate evaluation of cardiac risk before surgery is essential.
The findings suggest that incorporating AI ECG analysis into preoperative assessment could reduce unnecessary cardiac testing and associated healthcare costs.
"Considering the low cost and simplicity of ECG testing, this technology has significant potential for routine clinical practice," the research team said.
The investigators emphasized that AI ECG analysis is not intended to entirely replace conventional preoperative cardiac evaluation.
A preoperative electrocardiogram (ECG) analyzed by AI can identify patients at high risk of postoperative death and help rule out those unlikely to need additional cardiac testing, according to a new study by researchers at Seoul National University Bundang Hospital (SNUBH).
From left, Professor Choi Hong-mi, resident physician Kim Ye-rin, Professors Cho Young-jin and Song In-ae. (Courtesy of SNUBH)
The research team, led by Professor Choi Hong-mi, resident physician Kim Ye-rin, Professor Cho Young-jin of the Department of Cardiology, and Professor Song In-ae of the Department of Anesthesiology and Pain Medicine, analyzed data from 46,000 non-cardiac surgeries performed between 2020 and 2021 to evaluate the predictive performance of AI-based ECG analysis and its ability to guide the use of advanced cardiac examinations.
An ECG records the heart's electrical activity, but subtle waveform variations and complex patterns are often difficult for clinicians to interpret.
Building on this limitation, Professor Kim Joong-hee of the Department of Emergency Medicine and Professor Cho previously developed ECG Buddy, an AI-powered ECG analysis solution that quantifies hidden cardiovascular risks beyond conventional interpretation. The platform is now widely used in emergency departments across Korea.
In the latest study, the researchers applied multiple AI-derived cardiac biomarkers generated by ECG Buddy to preoperative risk assessment. Because anesthesia and surgery place additional stress on the cardiovascular system, accurate evaluation of cardiac risk before surgery is essential.
The team first examined the predictive value of the AI Critical Score (QCG-Critical Score), which estimates the likelihood of severe adverse events. The analysis found a strong correlation between the AI score and 30-day postoperative mortality. Patients with scores below 10 had a mortality rate of just 0.1 percent, while those with scores above 40 had a mortality rate of 11.7 percent.
The AI model achieved an area under the receiver operating characteristic curve (AUROC) of 0.909 for predicting 30-day postoperative mortality. This outperformed established international assessment tools, including the European Society of Cardiology's risk assessment method (AUROC 0.728), the Revised Cardiac Risk Index (RCRI; AUROC 0.725), and the American Society of Anesthesiologists (ASA) Physical Status Classification (AUROC 0.886).
The researchers also evaluated whether AI ECG analysis could identify patients who truly required advanced preoperative cardiac testing, such as echocardiography or coronary computed tomography (CT).
Using eight AI-derived ECG biomarkers associated with cardiac function and disease, patients were classified as either low risk, with all biomarkers within the normal range, or high risk if one or more biomarkers were abnormal.
The results showed that 92.3 percent of all surgical patients fell into the low-risk category. Among these patients, only 0.2 percent died or required emergency coronary intervention within 30 days after surgery. These findings suggest that many patients may not need advanced preoperative cardiac examinations, despite such tests accounting for 62.8 percent of the total cost of preoperative cardiovascular testing.
The findings suggest that incorporating AI ECG analysis into preoperative assessment could reduce unnecessary cardiac testing and associated healthcare costs.
Notably, all of the AI predictions were generated solely from a single ECG obtained before surgery, without requiring additional laboratory values or complex clinical data.
"Considering the low cost and simplicity of ECG testing, this technology has significant potential for routine clinical practice," the research team said.
The investigators emphasized that AI ECG analysis is not intended to entirely replace conventional preoperative cardiac evaluation. Rather, it should serve as a screening tool to identify high-risk patients while reducing the burden of unnecessary testing.
"AI-based ECG analysis is already widely used in emergency settings as a screening tool before more comprehensive examinations because it is fast, simple, and cost-effective," Professor Cho said. "Our findings suggest that it can also be applied in routine surgical care to reduce unnecessary testing while identifying patients at higher cardiovascular risk."
