This AI can predict sudden cardiac arrest—even in young people
What's the story
A new artificial intelligence (AI) model has proven to be more effective than doctors in predicting the risk of sudden cardiac arrest in patients with hypertrophic cardiomyopathy (HCM). This model analyzes electrocardiograms (ECGs) using deep learning and processes patient profiles to identify high-risk individuals before the event occurs. The model, named Multimodal AI for ventricular Arrhythmia Risk Stratification (MAARS), was developed by a team led by Dr. Natalia Trayanova at Johns Hopkins University.
Model performance
MAARS model more than twice as sensitive than current guidelines
The MAARS model predicts the risk of cardiac arrest by analyzing a wide range of medical data, including echocardiogram and radiology reports, as well as contrast-enhanced MRI (CMR) images of the patient's heart. Compared to current clinical guidelines, which only identify half of those who go on to experience cardiac arrest, MAARS was nearly twice as sensitive. It achieved an accuracy rate of 89% across all patients and 93% for those aged between 40-60 years.
Data significance
AI can analyze patterns in data humans might miss
The inclusion of CMR data is critical to the MAARS model, as it can detect scarring on the heart that characterizes hypertrophic cardiomyopathy. However, clinicians have struggled to use these images effectively due to the challenge of linking scar tissue patterns with clinical outcomes. Deep neural networks, like those used in MAARS, are particularly good at recognizing and analyzing such patterns in data that humans might miss.
Development details
HCM leading cause of sudden cardiac death in young individuals
The MAARS model was trained on data from 553 patients in The Johns Hopkins Hospital's hypertrophic cardiomyopathy registry. It was then tested on an independent external cohort of 286 patients. Hypertrophic cardiomyopathy is among the most common inherited heart conditions, affecting approximately 1 in 200 to 500 people worldwide. It is a leading cause of sudden cardiac death in young individuals and athletes, yet accurately predicting who is at risk of cardiac arrest remains a significant challenge.
Clinical implications
Model's predictions can help improve patient care
The MAARS model's ability to accurately predict the risk of serious adverse outcomes could greatly improve patient care. It would ensure that patients receive appropriate treatments to lower their risk, while avoiding unnecessary ones. This is particularly important in cases where implantable defibrillators are used as a precaution against sudden cardiac arrest, given the potential risks associated with the procedure.
Treatment customization
Need for rigorous external validation before clinical use
The MAARS model could also be used to customize treatment for patients with hypertrophic cardiomyopathy. It can pinpoint the most important parameters for each patient, which could help in managing their condition more effectively. However, experts caution that before MAARS can be widely adopted in clinical practice, it needs rigorous external validation across different institutions and healthcare settings.