Assessing how well the heart relaxes and fills with blood (diastolic function) is important for diagnosing certain heart conditions. When the heart does not relax properly, pressure builds up. Detecting this early is key, especially in patients with heart failure but normal pumping capacity. Typically, echocardiograms (heart ultrasounds) are used to evaluate diastolic function. However, these tests require extensive training and often give inconclusive results. About half of patients end up with unclear diagnoses.
Instead, artificial intelligence (AI) analysis of electrocardiograms (ECGs) shows promise. ECGs, which measure heart electrical signals using electrodes on the skin, are low-cost and widely available. In a study published by researchers, Eunjung Lee, Saki Ito, William R. Miranda, Francisco Lopez-Jimenez, Garvan C. Kane, Samuel J. Asirvatham, Peter A. Noseworthy, Paul A. Friedman, Barry A. Borlaug, Zachi I. Attia & Jae K. Oh from Department of Cardiovascular Medicine, Mayo Clinic, Rochester, have developed an AI model to predict diastolic dysfunction from 12-lead ECG results alone. We trained the model on nearly 100,000 patient ECGs with matched echocardiogram diagnoses.
Impressively, the AI model was able to detect abnormal relaxation and elevated heart filling pressures. It did so even in patients where the echocardiogram read as inconclusive. The AI model also predicted outcomes – patients it flagged with high pressures had higher long-term mortality.
In the future, this AI-powered ECG analysis could serve as an easily accessible test for cardiac diastolic issues. By enabling earlier detection, this approach has potential to improve outcomes for multiple heart disorder patients. The ability to leverage existing, non-invasive tools showcases the doors AI is opening for advancing healthcare.
In summary, it establishes that AI-enabled ECG is a simple yet powerful tool for identifying increased filling pressure and diastolic function grades. With a prognostic value comparable to echocardiography, AI-ECG emerges as a promising avenue to enhance the detection of diseases associated with diastolic dysfunction and increased diastolic filling pressure. As we delve deeper into the realm of AI applications in healthcare, the integration of such innovative technologies holds immense potential for transforming cardiac diagnostics and prognostics.