Can selfies help detect heart disease?
An artificial intelligence algorithm has been developed to safely use facial data and medical records to predict heart disease, a new study claims
Imagine that the next time you have to visit a cardiologist, he asks you to send a selfie instead of your past health records. As far as medical diagnostics go, this might sound strange. But according to a new study, doctors may now be able to detect heart diseases in a person with the help of their selfies.
The authors of the study, published in the European Heart Journal on 21 August, have shown that it may be possible to use a deep-learning algorithm to detect coronary artery disease — or CAD — by analysing four photographs of a person's face.
Researchers believe this algorithm has the potential to be deployed as a screening tool to identify possible heart disease in people both in the general population and high-risk groups, who could be then referred for further clinical investigations. “To our knowledge, this is the first work demonstrating that artificial intelligence can be used to analyse faces to detect heart disease. It is a step towards the development of a deep learning-based tool that could be used to assess the risk of heart disease, either in outpatient clinics or by means of patients taking ‘selfies’ to perform their own screening," said Professor Zhe Zheng, vice director of the National Center for Cardiovascular Diseases and vice-president of Fuwai Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, in Beijing, China.
Zheng, who led the research, explains that the ultimate goal is to develop a self-reported application for high-risk communities to assess heart disease risk before visiting a clinic. “This could be a cheap, simple and effective way of identifying patients who need further investigation," Zheng adds in an official release European Society of Cardiology’s website.
There are some known facial features that are associated with an increased risk of heart disease: including thinning or grey hair, ear lobe crease, wrinkles, xanthelasmata (small deposits of cholesterol underneath the skin, usually around our eyelids) and arcus corneae (fat, cholesterol deposits that appear as a hazy white, grey or blue opaque ring in the outer edges of the cornea). However, these features are difficult for humans to use to predict and quantify the risk of heart disease, the release adds.
For the research, Zheng and his colleagues enrolled 5,796 patients from eight Chinese hospitals and studied them between July 2017-March 2019. They were divided into two sections: a training group (5,216 patients, 90%) and a validation group (580, 10%). All these patients were also undergoing imaging procedures, such as coronary angiography or coronary computed tomography angiography, to investigate their blood vessels. Apart from this, research nurses used digital cameras to take four photographs of the patients — one frontal, two profiles and one view of the top of the head. Further data was compiled by interviewing the patients about their medical histories, lifestyle habits and socioeconomic status.
Radiologists also analysed angiograms and assessed the degree of heart disease in these patients, depending on how many blood vessels were narrowed by 50% or more and their location, the release adds. All this collection of data was then used to train and validate the computer algorithm.
Once the algorithm was ready, it was tested on a further 1,013 patients from nine hospitals across China, enrolled between April-July 2019. The researchers found that it out-performed two existing methods that predict heart disease risk: the Diamond-Forrester model and the CAD consortium clinical score. “In the validation group of patients, the algorithm correctly detected heart disease in 80% of cases (the true positive rate or 'sensitivity') and correctly detected that heart disease was not present in 61% of cases (the true negative rate or ‘specificity’), the release explains. In the test group, the ‘sensitivity’ was 80% and ‘specificity’ was 54%.
Despite this positive success, the algorithm does need to undergo further levels of refinement, the researchers added. There are also some other limitations and ethical concerns that need to be addressed before this tool is deployed among the general population. The algorithm not only needs to be tested in a larger group of people, from different ethnic backgrounds, but it also raises some questions over the potential misuse of personal information, especially facial data and records. These concerns were raised by fellow cardiovascular researchers from the University of Oxford in an accompanying editorial with the study.
Zheng admits in the release that ethical issues in developing and applying such novel technologies are of “key importance". “We believe that future research on clinical tools should pay attention to the privacy, insurance and other social implications to ensure that the tool is used only for medical purposes," he adds.
FIRST PUBLISHED23.08.2020 | 12:00 PM IST