Related Experiment Video
Updated: Jun 13, 2025
10:17
Real-Time Cardiac Mapping with a Noninvasive Imageless Electrocardiographic Imaging System
Published on: April 11, 2025
434
Deep behavioural representation learning reveals risk profiles for malignant ventricular arrhythmias
Maarten Z H Kolk1,2, Diana My Frodi3, Joss Langford4,5
1Department of Clinical and Experimental Cardiology, Amsterdam UMC Location University of Amsterdam, Heart Center, Meibergdreef 9, Amsterdam, the Netherlands.
NPJ Digital Medicine
|September 16, 2024
Summary
Deep learning identified five patient behavioral profiles from wearable data. Patients with low activity and poor sleep had a significantly higher risk of ventricular arrhythmias, informing personalized prevention strategies.
Area of Science:
- Cardiology
- Digital Health
- Machine Learning
Background:
- Sudden cardiac death (SCD) remains a significant concern in patients with high risk.
- Current risk stratification relies heavily on clinical factors, potentially missing behavioral insights.
- Wearable technology offers continuous monitoring of daily activities and sleep patterns.
Purpose of the Study:
- To identify and characterize distinct behavioral profiles in high-risk SCD patients using deep representation learning.
- To correlate these behavioral profiles with the risk of malignant ventricular arrhythmias.
- To explore the potential for personalized prevention strategies based on identified behaviors.
Main Methods:
- A prospective, observational study (SafeHeart) involving 272 patients receiving an implantable cardioverter-defibrillator (ICD).
- Unsupervised clustering was applied to low-dimensional representations of 180-day behavioral time-series data from wearable accelerometers, learned via a convolutional residual variational neural network (ResNet-VAE).
- Behavioral data from 37,478 days were analyzed to identify distinct patient clusters.
Main Results:
- Five distinct behavioral profiles were identified: low activity/poor sleep (Cluster A), moderate activity (Cluster B), high activity (Cluster C), good sleep (Cluster D), and poor sleep (Cluster E).
- Annual risks of malignant ventricular arrhythmias varied significantly, with Cluster A showing the highest risk (30.4%).
- Cluster A demonstrated a 3.63-fold increased risk of malignant ventricular arrhythmias compared to low-risk profiles (Clusters D-E), even after adjusting for clinical covariates (aHR 3.63, p < 0.001).
Conclusions:
- Deep representation learning can effectively characterize behavioral profiles in high-risk SCD patients from wearable data.
- Specific behavioral patterns, particularly low physical activity and disturbed sleep, are associated with a substantially elevated risk of malignant ventricular arrhythmias.
- These data support the development of personalized approaches for preventing ventricular arrhythmias and SCD, integrating behavioral insights into risk assessment.
Related Concept Videos
01:28Disturbances in Heart Rhythm
918
Arrhythmia or dysrhythmia refers to an abnormal heart rhythm caused by a defect in the heart's conduction system. It can cause the heart to beat irregularly, too quickly, or too slowly, leading to symptoms like chest pain, shortness of breath, and fainting. Factors such as stress, caffeine, alcohol, nicotine, cocaine, certain drugs, congenital defects, diseases, and electrolyte abnormalities can trigger arrhythmias.
Arrhythmias are categorized by their speed, rhythm, and origin. A slow...
Arrhythmias are categorized by their speed, rhythm, and origin. A slow...
918
01:28Mechanism of Cardiac Arrhythmias
900
Arrhythmias are irregular heart rhythms occurring when the heart's electrical impulses become abnormal. These disturbances can lead to various symptoms, depending on their severity and the underlying cause. Some common factors contributing to arrhythmias include hypoxia, ischemia, electrolyte imbalances, excessive catecholamine exposure, drug toxicity, and muscle overstretching. Arrhythmias can be classified into two main types based on the rate and site of origin of abnormal heart rhythms.
900