Machine learning could help predict which patients will develop diabetes, according to researchers in Japan.
Diabetes is linked to increased risks of severe health problems, including heart disease and cancer. Preventing diabetes is essential to reduce the risk of illness and death.
"Currently we do not have sufficient methods for predicting which generally healthy individuals will develop diabetes," said lead author Akihiro Nomura, M.D., Ph.D., of the Kanazawa University Graduate School of Medical Sciences in Kanazawa, Japan. "With machine learning, it could be possible to precisely identify high-risk groups of future diabetes patients better than using existing risk scores. In addition, the rate of visits to healthcare providers might be improved to prevent future onset of diabetes."
Nomura and colleagues analysed 509,153 nationwide annual health checkup records from 139,225 participants from 2008 to 2018 in the city of Kanazawa. Among them, 65,505 participants without diabetes were included.
The data included physical exams, blood and urine tests and participant questionnaires. Patients without diabetes at the beginning of the study who underwent more than two annual health checkups during this period were included. New cases of diabetes were recorded during patients' checkups.
The researchers identified a total of 4,696 new diabetes patients (7.2%) in the study period. Their trained computer model predicted the future incidence of diabetes with an overall accuracy of 94.9%.
Nomura says he next plans to perform clinical trials to assess the effectiveness of using statins to treat groups of patients identified by the machine learning model as being at high risk of developing diabetes.
The findings were published in an abstract to appear in the Journal of the Endocrine Society. The Endocrine Society canceled its annual meeting, ENDO 2020, amid concerns about COVID-19. However, accepted abstracts are now being made available online and will be published in a special supplemental section of the Journal.
