1552-P: Machine Learning Approach to Decision-Making for Initial Insulin Use in Japanese Patients with Type 2 Diabetes

Abstract

We examined whether the machine learning approach could be used as support for decision-making by general physicians as to whether to use insulin as an initial therapy for patients with type 2 diabetes. Prescription data on different antidiabetic agents from December 2009 to March 2015 from diabetes specialists’ patient registries were used to compare the machine learning decision of initial insulin use to that of general physicians. Gold standard was defined as consensus, that is, 80% agreement among nine diabetes specialists based on patients’ information. Twenty-two general physicians chose the most suitable medication according to the patients’ information. Analyzed were data on 4,860 patients who received initial monotherapy with either insulin (293 patients) or non-insulin (4,567 patients) and had laboratory data. Accuracy was calculated by 5-fold cross validation in machine learning. The average predictive values of initial insulin use by machine learning were 0.40, 0.50 and 0.82, respectively, for under-sampling ratios of 1:2, 1:4 and 1:8. General physicians achieved a mean accuracy for initial insulin use of 51%. The overall accuracy was 43%, 57% and 86% for under-sampling ratios of 1:2, 1:4 and 1:8. The difference in the accuracy between general physicians and machine learning in the under-sampling ratio of 1:8 was significant (P<0.05). In conclusion, accuracy of machine learning with the sampling ratio of 1:8 was higher compared with that of general physicians in initial insulin use defined by diabetes specialists’ choice as the gold standard. Assistance by machine learning may be beneficial to any general physician in deciding upon initial insulin usage. Disclosure K. Fujihara: None. M.H. Yamada: None. Y. Matsubayashi: None. M. Yamamoto: None. T. Iizuka: None. K. Miyamura: None. Y. Hasegawa: None. T. Yamazaki: None. S. Kodama: None. H. Sone: Research Support; Self; Kyowa Hakko Kirin Co., Ltd., Novartis AG, Ono Pharmaceutical Co., Ltd., Taisho Pharmaceutical Co., Ltd., Takeda Pharmaceutical Co. Funding Japan Society for the Promotion of Science (18K17897)