Prediction of incident atrial fibrillation: a comprehensive evaluation of conventional and AI-enhanced approaches.

Background: Multiple risk scores and biomarkers have been proposed for the prediction of atrial fibrillation (AF), but it is unknown how these compare to each other and if they could be combined.

Objective: Evaluate and compare approaches for incident AF prediction METHODS: The artificial intelligence-enhanced electrocardiogram (AI-ECG) Risk Estimator-AF (AIRE-AF), a convolutional neural network with a discrete-time survival loss function, was developed to predict incident AF. It was trained using a dataset of 1,163,401 ECGs from 189,539 patients from the Beth Israel Deaconess Medical Center (BIDMC) and externally validated in the UK Biobank (UKB, n = 38,892). AIRE-AF was compared to other risk prediction approaches including CHARGE-AF, a clinical risk score.

Results: In the BIDMC cohort, AIRE-AF predicted incident atrial fibrillation with a C-index of 0.750 (0.743-0.758). AIRE-AF was superior to CHARGE-AF, left atrial (LA size) and N-terminal pro-B-type natriuretic peptide (NTproBNP). The addition of CHARGE-AF and LA size provided a minor improvement in performance, (C-index improvement 0.017). There was no additive value of NTproBNP in combination with AIRE-AF. The single best performing single predictor in the volunteer population (UKB) was CHARGE-AF (C-index 0.750 (0.734-0.769)). The best performing combination of two predictors were AIRE-AF and CHARGE-AF, C-index 0.768 (0.743-0.792). The addition of a polygenic risk score (PRS) to AIRE-AF and CHARGE-AF provided a further significant improvement in performance (C-index 0.791 (0.766-0.816).

Conclusion: We present the first comprehensive evaluation of methodologies for predicting incident AF. Risk prediction with a model including AIRE-AF and CHARGE-AF resulted in similar performance to more complex models.