Algorithm detects heart disease by analyzing selfies

0

Researchers at the National Cardiovascular Disease Center in Beijing, China, have developed a method of detecting cardiac conditions that, according to them, could make the identification of patients at risk inexpensive, simple and effective. Using artificial intelligence and deep learning techniques, scientists gave birth to an algorithm capable of analyzing selfies and indicating possible problems.

Zhe Zheng, deputy director of the institution and main author of the article published in the European Heart Journal, celebrates the achievement, which, according to him, is a pioneer. The team involved hopes that the novelty can be used to make life easier for people, who would only need to take pictures and send them to clinics before making appointments.

Despite expectations, there is still a lot to do. “The algorithm requires more refinement and external validation in other populations and ethnicities,” explains Zheng. That’s because the tool looks for specific facial features, including thinning or auditory wear, wrinkles, creases on the earlobe, yellow deposits around the eyelids and cloudy white rings on the outer edges of the cornea – which can vary greatly from one patient to another .

Successful tests

It is no wonder that researchers are so hopeful. Surpassing existing methods, the study accurately detected 80% of heart disease cases and reported the absence of conditions with a 61% hit rate. More than 1,000 patients from nine Chinese hospitals have undergone evaluation.

Xiang-Yang Ji, from the Department of Automation at Tsinghua University and co-author of the article, points out that the number of false positives is one of the main issues to be addressed. “This can cause anxiety and inconvenience for patients, as well as potentially overloading clinics with people who require unnecessary screening.”

See Also
Algorithm that Reduces Human Intervention in Learning Robot from Google

Finally, Charalambos Antoniades, professor of cardiovascular medicine at the University of Oxford, does not fail to explain how positive the improvement of such a tool can be shown: “It can be highly relevant for regions of the globe that are underfunded and have weak screening programs. for cardiovascular diseases “, he declares.


LEAVE A REPLY

Please enter your comment!
Please enter your name here