Researchers at the Mayo Clinic have demonstrated that artificial intelligence (AI) can extract functional and physiological data from routine coronary angiography and predict four key cardiovascular biomarkers with over 80% accuracy. Mohamad Alkhouli, MD, from the Mayo Clinic Alix School of Medicine, shared these findings at the Society for Cardiovascular Angiography and Intervention (SCAI) 2024 scientific sessions.
Alkhouli explained that the AI-ENCODE study utilized advanced machine-learning techniques to analyze data from 20,000 angiograms performed at the Mayo Clinic. The AI algorithms were trained to assess left and right ventricular functions, intracardiac filling pressures, and cardiac index, using echocardiography and right heart catheterization as validation comparators.
The AI models accurately predicted left ventricular ejection fraction, left ventricular filling pressures, right ventricular dysfunction, and cardiac output with area under the curve scores of 0.87, 0.87, 0.80, and 0.82, respectively. Alkhouli noted that while the algorithms still require refinement, the ultimate aim is to integrate these predictive models into a real-time dashboard for use in catheterization labs.
As reported by Medscape, future plans include developing AI algorithms to predict heart valve calcium, pericardial restriction, transplant rejection, and regional wall motion abnormalities, which will involve data cleaning, external validation, and enhancing IT infrastructure.
Ian Gilchrist, MD, from the Milton S. Hershey Medical Center, expressed cautious optimism about the potential of AI in medicine, noting that widespread adoption will depend on demonstrating cost-effectiveness and ensuring robust system interconnectivity.
Alkhouli emphasized that AI could augment clinical practice by allowing physicians to focus on procedures while simultaneously providing critical supplementary data. He reassured that AI is a tool to enhance human capability, not a replacement, encouraging intelligent integration to maximize its benefits.