An artificial intelligence (AI) tool designed to help doctors identify high-risk patients for heart disease is set to undergo trial in England. A study revealed its ability to accurately predict mortality risk.
The international research team, led by Imperial College London, trained the AI model, known as AI-ECG risk estimation (AIRE), using millions of electrocardiogram (ECG) results. This common medical test records electrical signals within and between the heart's chambers, typically used to diagnose heart diseases and anomalies.
The objective is to identify nuanced patterns that may indicate a person is at high risk for health issues or mortality. In trials, the model predicts the likelihood of death over a decade following an ECG, achieving an accuracy rate of 78%.
Dr. Fu Siong Ng, a cardiac electrophysiology researcher at Imperial College London, noted the potential benefits for the National Health Service (NHS) and globally. The system can also predict heart attacks, heart failure, and rhythm issues, with researchers indicating it could be implemented within the NHS in the next five years.
Real patient trials are planned at several locations in London, expected to commence by mid-2025, to assess the model's benefits through outpatient and hospital patient data.
AI-driven ECGs are already utilized for diagnosing heart diseases, but they are not part of routine medical care and have not been employed to determine individual patient risk levels. Brian Williams, Chief Scientific and Medical Officer at the British Heart Foundation, which funded the study, explained that this could extend ECG usage beyond current capabilities, aiding in assessing future heart and health problem risks, as well as mortality risk.
Researchers, who published their findings in the journal Lancet Digital Health, acknowledged that errors in AI predictions could stem from unknown factors, such as additional treatments received by patients or unexpected deaths. Nonetheless, the model may capture subtle changes in heart structure that serve as warning signs for disease or mortality risk, which doctors might overlook.
Dr. Arunashis Sau, an academic physician at Imperial College London leading the new research, emphasized that while cardiologists rely on experience and standard guidelines to interpret ECGs, the AI model detects finer details, potentially identifying issues in ECGs that appear normal and doing so well before diseases fully develop.
Further research in hospitals and other health facilities is necessary to determine the model's future role in diagnostics and treatment, with potential benefits for patients with other health issues, as conditions like diabetes can also impact heart health.