Generative AI Technology Detects Heart Disease Earlier Than Conventional Methods

Detecting heart dysfunction early using cost-effective and widely accessible tools like electrocardiograms (ECGs) and efficiently directing the right patients for more expensive imaging tests remains a significant challenge in healthcare. Researchers have now developed an artificial intelligence (AI) technology that transforms basic ECG readings of electrical activity into advanced heart motion signals, typically obtained through echocardiograms, offering the potential to improve the detection and monitoring of heart disease.

This innovative, patent-pending technology was developed by researchers at Rutgers Health (Newark, NJ, USA) and RWJBarnabas Health (West Orange, NJ, USA). It utilizes generative AI to analyze the speed of cardiac tissue movement during a heartbeat, using electrical signals from an ECG. The AI then converts these signals into a speed waveform, resembling the waves traditionally measured through Doppler ultrasound during echocardiography. These waveforms allow doctors to assess how well the heart is contracting and relaxing during each heartbeat. By employing generative adversarial networks (GANs), the team trained AI models to produce synthetic heart motion waveforms from electrical signals. Extensive testing conducted across various clinical sites in the U.S. and Canada demonstrated the technology’s high accuracy in detecting both diastolic dysfunction (issues with heart relaxation) and systolic dysfunction (issues with heart contraction). This approach allows for earlier detection of heart problems compared to conventional methods.

Following the development of the system, the team took multiple steps to validate the synthetic heart-motion waveforms generated from ECG data. In a randomized test, board-certified echocardiographers were unable to distinguish between real and AI-generated waveforms. The synthetic measurements also reflected patient-specific variations in physiology and clinical factors such as age and blood pressure, just as real echocardiographic measurements would. Most notably, synthetic tissue Doppler imaging (TDI) successfully predicted clinical outcomes. The innovation could lead to earlier identification of heart disease and reduce unnecessary medical tests.

Analysis revealed that the synthetic TDI technique could decrease the need for echocardiograms by 64.3% for detecting left ventricular systolic dysfunction and by 69.9% for detecting diastolic dysfunction, with only a 1.4% and 6.5% miss rate, respectively. Beyond its role in screening, the technology has broader clinical applications. It could improve the monitoring of cancer patients undergoing cardiotoxic therapies or patients with hypertrophic cardiomyopathy taking new medications that impact heart muscle function. Looking to the future, the researchers envision a system with digital “twins” of patients' hearts, allowing doctors to test potential treatments virtually on the digital twin before making decisions for the actual patient.

“When risk factors like high blood pressure, diabetes or coronary artery disease start affecting the heart muscle, conventional measurements in ECG-detectable changes come very late,” said senior study author Partho Sengupta. “The synthetic TDI can detect subtle longitudinal changes in heart function long before the pumping fraction decreases. The novelty is not just relying on a change in pumping fraction – also referred to as ejection fraction – but more subtle changes in heart tissue motion.”

Related Links:
Rutgers Health
RWJBarnabas Health