A study that can predict in advance whether a person who has been pointed out as having diabetes in a specific medical examination will actually see a medical institution from four items such as the frequency of seeing a medical institution in the past and the type of medicine used. The results were announced by groups such as the University of Tokyo.
This research was conducted by a group of Associate Professor Satoko Yamaguchi of the University of Tokyo Hospital.
In health examinations such as specific medical examinations, it is pointed out that diabetes is present, and even if it is recommended to see a medical institution, it is said that only about one-third of the actual medical examinations are actually done, which is an issue.
The group used a method called "machine learning" to determine whether or not more than 10,000 people who meet the diagnostic criteria for diabetes registered with the medical statistics database company will be examined within half a year. I analyzed it.
As a result, the tendency not to see a doctor was related to the fact that the frequency of seeing a medical institution in the past year was low, the level of "HbA1c" related to blood sugar level was low, and the drug for lipid abnormality was prescribed. There were four items: no medicine and no prescription for hypertension.
According to the group, by examining these four items, it is possible to predict the patient's consultation behavior after the examination to a certain extent, and Associate Professor Yamaguchi said, "Since it can be predicted with only a few items, we will create a mechanism to efficiently connect to medical institutions. I think it will be useful for that. "