Assessment of causal association between ECG aging predicted by artificial intelligence and the risk of atrial fibrillation: a Mendelian randomization analysis

Artificial intelligence (AI)-derived electrocardiographic aging (ECG-aging) has been found to be associated with the risk of atrial fibrillation (AF). We aimed to assess the causal association between the discrepancy in AI-predicted electrocardiographic age and chronological age (AI-ECG age gap) and AF risk.

In UK Biobank participants, AI-ECG age gaps was derived from 12-lead ECGs using a publicly available AI-ECG age prediction model. Both measured and genetically predicted AI-ECG age gaps were tested for association with incident AF using multivariable regression. Causal estimates for the Mendelian randomization (MR) analysis of the AI-ECG age gap and AF were derived using both individual-level data with a genetic risk score based on specific genetic variants, and summary-level data from genome-wide association studies. Inverse variance weighted was the primary MR method for causality, with additional sensitivity analyses used to test robustness.

In observational and linear MR analyses, each 1-SD increase in AI-ECG age gaps was associated with an increased risk of AF in both measured (HR 1.36 [95% CI, 1.22–1.53]) and genetically predicted (OR 1.06 [95% CI, 1.01–1.12]) age gaps. Two-sample MR results using two independent large datasets replicated the effect (OR 1.13 [95% CI, 1.02–1.25] and OR 1.13 [95% CI, 1.02–1.25], respectively), with sensitivity analyses confirming the robustness of the main findings. Non-linear MR analyses demonstrated a consistent and robust causal association.

This study provides evidence supporting a causal association between ECG-aging and the risk of AF. Our findings highlight ECG-aging as a potential causal mediator in the development of AF, helping to identify individuals at risk.