A joint research team from Imperial College London and Imperial College Healthcare NHS Trust has published breakthrough results in the European Heart Journal, developing an artificial intelligence algorithm capable of early prediction of heart valve disease from electrocardiograms (ECGs). This international collaborative study, involving institutions including Zhongshan Hospital in Shanghai, China, trained the model using nearly one million ECG records.

The AI system developed by the team can detect subtle changes in cardiac electrical activity from routine ECGs, predicting the risk of mitral, tricuspid, or aortic valve regurgitation years before structural abnormalities are detectable by echocardiography. Individuals flagged as "high-risk" by the algorithm are 10 times more likely to develop severe heart valve problems than low-risk individuals.

"This technology allows us to identify high-risk individuals before symptoms appear," said lead researcher Dr. Arunashis Sau. After training on over 400,000 patient records in China, the AI model validated its cross-racial applicability in more than 34,000 patients in the United States. The study showed the algorithm accurately predicted 69%–79% of heart valve disease cases.

Heart valve disease affects approximately 41 million people worldwide, and early diagnosis is critical for treatment. This AI technology provides clinicians with a new screening tool, potentially transforming current diagnostic models that rely on symptom onset or costly imaging. The team plans to further refine the algorithm and expand its application in early warning for cardiovascular diseases.