The heart health status is understood through retinal scanning (schematic diagram). Source: Official Website of Leeds University According to the latest issue of nature machine intelligence, British researchers have developed an artificial intelligence (AI) system that can identify patients at high risk of heart attack by analyzing eye scan data left during routine visits to optical shops or hospital ophthalmology. The recognition accuracy of the AI system is between 70% and 80%, which can be used as the second referral mechanism for cardiovascular disease screening. Changes in retinal microvessels are an indicator of a wider range of vascular diseases, including heart problems. In the study led by the University of Leeds in the UK, researchers used deep learning technology to train AI systems to automatically read retinal scanning data and identify those who may have heart disease in the next year. Deep learning is a series of complex algorithms that enable computers to recognize patterns in data and make predictions. Alex Frankie, a professor of computational medicine at the University of Leeds who led the study, said: "this technology has the potential to revolutionize heart disease screening. Retinal scans are relatively cheap and are often used in many matching services. As a result of automatic screening, people at high risk can be referred to a specialist for treatment." British biobank provided data for the study. In the process of deep learning, the AI system analyzed the retinal scan and heart scan data of more than 5000 people. The AI system determined the association between retinopathy and cardiac changes in patients. Once the image mode is learned, the AI system can estimate the size and pumping efficiency of the left ventricle (one of the four chambers of the heart) only through retinal scanning. Ventricular enlargement is associated with an increased risk of heart disease. With information about the estimated size of the left ventricle and its pumping efficiency, as well as basic demographic data such as the patient's age and gender, the AI system can predict their risk of heart attack in the next 12 months. At present, only after diagnostic tests such as echocardiography or cardiac magnetic resonance imaging can we determine the details of the patient's left ventricular size and pumping efficiency. These diagnostic tests are usually expensive and can only be used in hospitals, which makes them unavailable to people in countries with less resources in the health care system, and increases the cost and waiting time of health care in developed countries. "The AI system is a great tool to unlock the complex patterns in nature, and the complex patterns of retinal changes we found related to heart changes are one of them," said Swain prien, a professor of cardiovascular imaging at the British Heart Foundation at the University of Leeds and one of the authors of the study (Xinhua News Agency)