Artificial intelligence-enhanced Electrocardiography Score for Perioperative Risk Assessment in Non-cardiac Surgery

Background

The role of electrocardiography (ECG) has been limited in the preoperative risk evaluation in noncardiac surgery due to its low prognostic value. Recent advances in artificial intelligence (AI) have enabled the extraction of subtle features from ECG that can be used in risk prediction.

Objectives

This study aimed to evaluate the utility of an AI-enabled ECG (QCG-Critical score) in predicting 30-day postoperative mortality in non-cardiac surgery and compare its performance with traditional perioperative risk-assessment tools.

Methods

A retrospective cohort of 46,135 non-cardiac surgeries at a tertiary center between 2020 and 2021 was analysed. Preoperative ECG images acquired within 30 days before surgery were input to a previously developed CNN-based deep-learning algorithm to generate QCG-Critical score, and its ability to predict 30-day in-hospital mortality was evaluated. Secondary outcomes included 7-day mortality, prolonged mechanical ventilation, unplanned percutaneous coronary intervention, and heart failure within 30 days postoperatively.

Results

The 30-day postoperative mortality rate was 0.34%. The QCG-Critical score model showed strong predictive performance (AUROC: 0.909) predicting postoperative mortality, outperforming the ESC guidelines (0.728) and RCRI (0.725) and performing comparably to the ASA classification (0.886). The QCG-Critical score model remained predictive across clinical subgroups and demonstrated superior performance in predicting secondary outcomes: 7-day mortality, unplanned percutaneous coronary intervention, mechanical ventilation, and heart failure.

Conclusion

The preoperative QCG-Critical score accurately predicted postoperative mortality and other adverse outcomes, outperforming conventional risk stratification tools.

The QCG-Critical score may serve as a fast, accessible, and integrable tool for perioperative risk assessments in routine surgical care.

Condensed Abstract

The AI-driven QCG-Critical score predicted 30-day postoperative mortality after non-cardiac surgery with excellent accuracy (AUROC 0.909), outperforming traditional risk models. It also showed strong predictive power for adverse outcomes including early mortality, unplanned PCI, and heart failure. Unlike existing models, the QCG-Critical score requires only ECG images and no additional clinical inputs, enabling rapid, objective, and scalable perioperative risk assessment. This AI-enabled approach may optimize surgical decision- making and resource utilization.