AI Technology Predicts Cardiac Events 10 Years in Advance

Current clinical guidelines recommend coronary computed tomography angiography (CCTA) as a first-line investigation for stable chest pain to identify patients who may need coronary revascularization due to obstructive coronary artery disease (CAD). However, this method often identifies many patients who do not have obstructive CAD, as well as those without any coronary atheroma, who are typically reassured and discharged without further treatment or follow-up, leaving their management and outcomes uncertain. Now, a landmark clinical study recently published in The Lancet has demonstrated the efficacy of a novel artificial intelligence (AI) technology in quantifying coronary artery inflammation and predicting cardiac events accurately.

Developed by Caristo Diagnostics (Oxford, UK), the AI-enabled CaRi-Heart technology introduces a new approach to combating heart attacks and other cardiac diseases by detecting hidden inflammation. This technology leverages advanced AI algorithms applied to routine CCTA scans, allowing it to visualize and quantify coronary inflammation, a key but often invisible factor contributing to many fatal heart attacks and strokes. The technology was tested in a landmark study analyzing data from the first 40,000 patients enrolled in the ORFAN registry, which is the largest global study assessing the ability of coronary CCTA imaging biomarkers to predict long-term cardiovascular outcomes.

The study revealed that over 80% of patients who underwent CCTA did not show obstructive CAD at the time of their scans. Despite this, the group without obstructive CAD experienced twice as many fatal and non-fatal cardiac events. Coronary inflammation, measured by Caristo’s CaRi-Heart FAI-Score, predicted these events—including heart attacks and new cases of heart failure—independently of traditional risk factors, routine clinical CCTA interpretations, calcium scoring, and plaque quantification, and it did so up to 10 years in advance. Notably, among those patients who showed no or minimal coronary plaque initially, those with the most abnormal FAI-Score results faced a 9.5-fold increase in risk for cardiac mortality and a 5.5-fold increase in risk for major adverse cardiac events (MACE).

Furthermore, Caristo’s AI-Risk model, the CaRi-Heart Risk Score, surpassed other existing clinical scores in predicting cardiac mortality. When this score was presented to clinicians, it influenced changes in management decisions for 45% of the patients, primarily driven by the need to address previously undetected coronary inflammation. This study underscores the need for robust risk prediction tools that can identify patients at risk due to inflamed coronary arteries, especially those without obstructive CAD. The CaRi-Heart technology could transform CCTA from merely a diagnostic test for selecting a minority of patients for further intervention into a preventive tool that guides the management of all patients undergoing CCTA.

"Coronary inflammation is a crucial piece of the puzzle in predicting heart attack risk,” said Keith Channon, MD, Professor of Cardiovascular Medicine at the University of Oxford, Caristo's Chief Medical Officer. “We are excited to discover that CaRi-Heart results performed exceptionally well in predicting patient cardiac events. This tool is well positioned to help clinicians identify high-risk patients with seemingly 'normal' CCTA scans."