Cardiovascular events and artificial intelligence-predicted age using 12-lead electrocardiograms

Area of Science:

• Cardiology
• Artificial Intelligence
• Biomedical Engineering

Background:

• Electrocardiograms (ECGs) show potential for predicting biological age, a factor linked to cardiovascular events.
• The effectiveness of artificial intelligence (AI) in predicting age from ECGs and its association with cardiovascular outcomes requires further investigation.

Purpose of the Study:

• To develop and validate an AI-enabled ECG model for predicting chronological age (CA).
• To assess the utility of AI-predicted age from ECGs in forecasting cardiovascular events compared to CA.

Main Methods:

• A convolutional neural network was used to develop an AI-enabled ECG model on a dataset of 17,042 sinus rhythm ECGs (SR-ECGs).
• The model predicted chronological age, and the difference between AI-predicted age and CA (AgeDiff) was analyzed.
• Receiver operating characteristic (ROC) curves were used to evaluate the predictive performance of AI-predicted age and CA for cardiovascular events.

Main Results:

• During a mean follow-up of 460.1 days, 543 cardiovascular events occurred.
• The incidence of cardiovascular events increased with greater age differences (AgeDiff).
• AI-predicted age demonstrated superior prediction of cardiovascular events compared to CA in younger patients (under 60 years), with AUCs of 0.700 vs. 0.642 (P=0.003). No significant difference was observed in older patients.

Conclusions:

• AI-predicted age derived from 12-lead ECGs offers improved prediction of cardiovascular events in younger individuals compared to chronological age.
• The predictive utility of AI-predicted age for cardiovascular events was not superior to chronological age in older individuals.