AI System Predicts Heart Failure Risk Years in Advance from Holter ECG

Heart failure, a progressive syndrome in which the heart cannot meet the body’s circulatory demands, affects about 64 million people worldwide. In developed countries, roughly 12% of adults older than 65 live with the condition, which causes fatigue, shortness of breath, edema, and exercise intolerance and can be fatal. Earlier identification before clinical onset could enable preventive treatment and reduce deterioration. To help address this challenge, researchers have developed an artificial intelligence system that predicts future heart failure risk from routine electrocardiogram data.

DeepHHF is an artificial intelligence model created at the Technion Faculty of Biomedical Engineering (Haifa, Israel). The system analyzes standard 24-hour ambulatory electrocardiogram (Holter ECG) recordings obtained during routine home Holter monitoring. It aims to flag patients at elevated risk of developing heart failure well ahead of symptom onset.

The model operates on raw Holter ECG signals and detects subtle abnormalities that are often imperceptible to the human eye. Its output provides an early warning that can precede a diagnosis by several years. By focusing on data already captured in usual care, the approach is designed to integrate with existing cardiac monitoring pathways.

In development, DeepHHF was trained on approximately 70,000 Holter examinations performed at Leumit Health Services. The study was led by the Technion and published in npj Digital Medicine on June 19, 2026. Collaborating institutions included Rambam Health Care Campus, Shaare Zedek Medical Center, the Hebrew University of Jerusalem, Leumit Health Services, and Hadassah Medical Center.

The research underscores a growing emphasis on early diagnosis to enable preventive interventions that can improve outcomes and save lives. Because the method relies on noninvasive, widely available monitoring, it may help clinicians identify high-risk patients earlier and plan proactive management. The authors report that the model can predict risk up to five years in advance using raw Holter ECG recordings.

“To the best of our knowledge, no existing model can predict the risk of heart failure up to five years in advance using raw Holter ECG recordings. By relying on standard, non-invasive diagnostic tools, our model provides clinically valuable information that enables early identification of high-risk patients and timely preventive interventions, with the potential to reduce hospitalizations, suffering and mortality,” said Prof. Joachim Behar, senior author at the Technion Faculty of Biomedical Engineering.

Related Links
Technion Faculty of Biomedical Engineering