The AI ECG solution, ECG Buddy, uses the QCG-Critical score to predict 30-day postoperative mortality, achieving an AUROC of 0.909 and outperforming standard models like ESC Risk Assessment, RCRI, and with performance comparable to ASA.

The AI system derives its risk from multiple AI-derived cardiac biomarkers, with higher QCG-Critical scores correlating with higher mortality (0.1% at scores <10 versus 11.7% at scores >40).

Findings were published in peer-reviewed journals, including European Heart Journal– Digital Health and the Journal of Medical Internet Research, validating the approach.

Eight AI-derived ECG biomarkers classified patients into low risk (all normal) or high risk (any abnormal), with 92.3% of patients in the low-risk group and only 0.2% of these patients experiencing death or emergency coronary intervention within 30 days.

Among low-risk patients, 91% of preoperative advanced tests were normal, and these tests represented 62.8% of total preoperative cardiovascular testing costs, indicating a substantial burden reduction.

Eight biomarkers helped determine when additional tests, such as echocardiography or coronary CT, are warranted, effectively distinguishing normal from multiple abnormal findings.

An international study from Seoul National University Bundang Hospital and Kim Ye-rin Hospital analyzed 46,000 non-heart surgeries (2020–2021) to assess whether AI-enabled ECG analysis can predict 30-day mortality and guide the need for further scrutiny.

The study evaluated AI-analyzed preoperative ECGs to predict 30-day mortality and guide advanced cardiac testing, using a large real-world surgical cohort.

Authors position AI ECG as a screening gate to identify high-risk patients and reduce testing burden, rather than a full replacement for existing preoperative testing.

Importantly, all assessments were based on a single preoperative ECG without laboratory data, underscoring the approach’s low cost and simplicity.

The risk gradient is clear: QCG-Critical scores below 10 correspond to 0.1% 30-day mortality, while scores above 40 associate with 11.7% mortality.