The ECG2HF AI model demonstrated strong predictive ability for 10-year incident heart failure, with AUROC values of 0.88, 0.86, and 0.86 across multiple test sets.
Does an ECG-based AI model (ECG2HF) improve prediction of incident heart failure compared to the PCP-HF score in ambulatory patients?
A publicly available 12-lead ECG-based AI model (ECG2HF) accurately predicts 10-year incident heart failure risk, outperforming traditional clinical risk scores.
Absolute Event Rate: 0% vs 0%
Abstract Background Heart failure (HF) is a major public health problem associated with substantial morbidity. ECG-based artificial intelligence (AI) may enable efficient identification of individuals at elevated risk for future HF and enable targeted risk factor modification and preventive efforts. Existing models are proprietary with modest or inconsistent performance. Purpose To develop and validate a generalizable and publicly available convolutional neural network to predict incident HF using the 12-lead ECG waveform (Electrocardiogram-to-Heart Failure, "ECG2HF"). Methods We developed ECG2HF in 100,117 patients receiving longitudinal ambulatory care in our hospital, and validated the model among individuals without HF in one internal test set and 2 external test sets. HF events at 10 years were identified using a validated natural language processing model which has been shown to ascertain HF events using electronic health records with clinical adjudication committee-level accuracy. Discrimination was quantified by calculating the area under the receiver operating characteristic curve (AUROC) as well as hazard ratios per quintile of ECG2HF risk. We also compared discrimination and net reclassification (at 10%, 10-20%, ≥20% 10-year risk categories) using ECG2HF versus the 15-component Pooled Cohort Equations to Prevent Heart Failure (PCP-HF) score. Results The test sets comprised MGH (16,351 individuals, 578 events, age 56±17, 51% women), BWH (72,770 individuals, 2,743 events, age 56±16, 56% women), and BIDMC (30,344 individuals, 1,244 events, age 57±17, 55% women). ECG2HF discriminated 10-year incident HF consistently across each test set (AUROC MGH 0.88 0.86-0.89; BWH: 0.86 0.85-0.87; BIDMC 0.86 0.85-0.87). Compared to PCP-HF, ECG2HF resulted in favorable discrimination (improvement in AUROC MGH/BWH 0.053 0.020-0.086; BIDMC 0.038 -0.0096-0.086), and net reclassification (NRI MGH/BWH 0.17 0.095-0.25; BIDMC 0.23 0.10-0.35) of 10-year HF risk. When compared to the lowest quintile, hazard ratios for incident HF were markedly higher in the highest quintile (MGH: HR 82.80 36.97-185.46, BWH: HR 84.04 57.47-122.89, BIDMC HR 102.7 55.0-191.6). Performance of ECG2HF remained consistent across subgroups of age, sex and the presence or absence of common cardiac risk factors. Conclusion ECG2HF is a publicly available 12-lead ECG-based AI model that discriminates risk of future HF with favorable and consistent performance across multiple ambulatory healthcare populations. ECG2HF may enable efficient prioritization of high-risk individuals for HF-related preventive measures using a single 12-lead ECG.Figure 1.
Kany et al. (Thu,) reported a other. The ECG2HF AI model demonstrated strong predictive ability for 10-year incident heart failure, with AUROC values of 0.88, 0.86, and 0.86 across multiple test sets.