Specialists at the Kanazawa University Graduate School of Medical Sciences (Japan) have developed a computer algorithm that will help identify patients at risk of developing diabetes. This was reported in the journal of the American Endocrinological Society.
Scientists using artificial intelligence analyzed about 510 thousand reports on health checks of 139 thousand residents of the city of Kawazawa, conducted from 2008 to 2018. Of these, approximately 65 thousand people at the time of the start of the study were not sick with diabetes.
The reports included physical examinations, blood and urine tests, and questionnaires. All cases of diabetes in previously healthy people were noted: a total of 4,696 or 7.2% of such patients over the decade studied. Artificial intelligence was able to predict the appearance of their disease with an accuracy of 94.9%.
“Machine learning is likely to help us identify groups of people with a potentially high risk of developing diabetes more accurately than using existing risk assessment parameters,” said project manager Dr. Akihiro Nomura.
Now, Japanese experts intend to develop a plan for the early treatment of people at high risk.