ECG Based Unsupervised Clustering Model Predicts Heart Failure and Cardiovascular Diseases in Individuals without Cardiovascular Disease

Early detection of cardiovascular risk in the general population is challenging.We hypothesize that individuals without known cardiovascular disease (CVD) but at high risk may share electrocardiographic (ECG) features suitable for non-invasive risk stratification.This study aimed to identify clusters of individuals without CVD based on ECG morphology using unsupervised clustering and assess their association with incident CVD risk.A median heartbeat was derived from 10-second 12lead resting ECGs from 56,251 individuals without prevalent CVD in the UK Biobank. An unsupervised model classified individuals into k distinct clusters based on their ECG features. Survival analysis assessed the association of each cluster with incident cardiac risk (4-years followup).The model distinguished 2 clusters with varying ECG features, which significantly differed in terms of HF and CVD rate.Cluster 2 (N = 25,097) included the highest rate of heart failure (HF) (0.5%, p<0.001) and CVD (4.2%, p<0.001).These individuals exhibited ECG features that are associated with higher cardiac risk.Our analysis identified a group of individuals at risk of HF and CVDs using 10-second ECGs enabling fast and noninvasive risk assessment in apparently healthy populations.