The DeepHHF deep learning model applied to 24-hour single-lead ECG recordings predicted incident heart failure within 5 years with an AUROC of 0.80, significantly outperforming the clinical PCP-HF score.
Observational (n=40,174)
Yes
Does a deep learning model applied to 24-hour single-lead ECG data improve the prediction of heart failure risk within five years compared to short-segment ECGs and clinical scores?
Deep learning applied to 24-hour single-lead Holter ECGs can accurately predict incident heart failure within five years, outperforming standard clinical scores and short-segment ECG models.
Effect estimate: AUROC 0.80
Absolute Event Rate: 0.8% vs 0.74%
p-value: p=<0.05
Heart failure (HF) affects 11.8% of adults aged 65 and older, reducing quality of life and longevity. Preventing HF can reduce morbidity and mortality. We hypothesized that artificial intelligence (AI) applied to 24-hour single-lead electrocardiogram (ECG) data could predict the risk of HF within five years. To research this, the Technion-Leumit Holter ECG (TLHE) dataset, including 69,663 recordings from 47,729 patients, collected over 20 years, was used. Our deep learning model, DeepHHF, trained on 24-hour ECG recordings, achieved an area under the receiver operating characteristic curve of 0.80 that outperformed a model using 30-second segments and a clinical score. High-risk individuals identified by DeepHHF had a two-fold chance of hospitalization or death incidents. Explainability analysis showed DeepHHF focused on arrhythmias and heart abnormalities. This study highlights the feasibility of deep learning to model 24-hour continuous ECG data, capturing paroxysmal events essential for reliable risk prediction. Artificial intelligence applied to single-lead Holter ECG is non-invasive, inexpensive, and widely accessible, making it a promising tool for HF risk prediction.
Published June 19, 2026; high social media discussion in cardiology circles; featured in Nature portfolio.
Zvuloni et al. (Fri,) conducted a observational in Heart failure risk (n=40,174). DeepHHF (Deep learning model on 24-hour single-lead ECG) vs. PCP-HF score was evaluated on Incident heart failure within 5 years (AUROC 0.80, p=<0.05). The DeepHHF deep learning model applied to 24-hour single-lead ECG recordings predicted incident heart failure within 5 years with an AUROC of 0.80, significantly outperforming the clinical PCP-HF score.