Search

Identification of medical conditions using claims data is generally conducted with algorithms based on subject-matter knowledge. However, these claims-based algorithms (CBAs) are highly dependent on the knowledge level and not necessarily optimized for target conditions. We investigated whether machine learning methods can supplement researchers' knowledge of target conditions in building CBAs. Retrospective cohort study using a claims database combined with annual health check-up results of employees' health insurance programs for fiscal year 2016-17 in Japan (study population for hypertension, N = 631,289; diabetes, N = 152,368; dyslipidemia, N = 614,434). We constructed CBAs with logistic regression, k-nearest neighbor, support vector machine, penalized logistic regression, tree-based model, and neural network for identifying patients with three common chronic conditions: hypertension, diabetes, and dyslipidemia. We then compared their association measures using a completely hold-out test set (25% of the study population). Among the test cohorts of 157,822, 38,092, and 153,608 enrollees for hypertension, diabetes, and dyslipidemia, 25.4%, 8.4%, and 38.7% of them had a diagnosis of the corresponding condition. The areas under the receiver operating characteristic curve (AUCs) of the logistic regression with/without subject-matter knowledge about the target condition were .923/.921 for hypertension, .957/.938 for diabetes, and .739/.747 for dyslipidemia. The logistic lasso, logistic elastic-net, and tree-based methods yielded AUCs comparable to those of the logistic regression with subject-matter knowledge: .923-.931 for hypertension; .958-.966 for diabetes; .747-.773 for dyslipidemia. We found that machine learning methods can attain AUCs comparable to the conventional knowledge-based method in building CBAs.

AIM:Little is known about whether and how local-level resources regarding home care are associated with the prevalence of home deaths. We aimed to investigate whether geographic patterns of the resources for home care were associated with the prevalence of home deaths, taking spatial variation into consideration. METHODS:We conducted an ecological cross-sectional study in Japan using nationwide data in 2014. The areal unit was the municipality, the smallest administrative unit in Japan. We investigated the association between the percentage of home deaths and the resources of home care support clinics with available 24-hour-a-day functions, considering the geographic effect of neighboring municipalities by applying a geographically weighted regression model. RESULTS:The mean and standard deviation of the percentages of home deaths were 11.4% (5.0%), and those of the number of home care support clinics per 10,000 elderly population were 3.4 (3.7). The percentages of home deaths in neighboring municipalities tended to be significantly correlated (Moran's I 0.34, p

AIMS:The purpose of this study was to examine whether periodontitis is associated with incident type 2 diabetes in a Japanese male worker cohort. METHODS:The study participants were Japanese men, aged 36-55 years, without diabetes. Data were extracted from the MY Health Up study, consisting of self-administered questionnaire surveys at baseline and following annual health examinations for an insurance company in Japan. The oral health status of the participants was classified by two self-reported indicators: (1) gingival hemorrhage and (2) tooth loosening. Type 2 diabetes incidence was determined by self-reporting or blood test data. Modified Poisson regression approach was used to estimate the relative risks and the 95% confidence intervals of incident diabetes with periodontitis. Covariates included age, body mass index, family history of diabetes, hypertension, current smoking habits, alcohol use, dyslipidemia, and exercise habits. RESULTS:Of the 2895 candidates identified at baseline in 2004, 2469 men were eligible for follow-up analysis, 133 of whom were diagnosed with diabetes during the 5-year follow-up period. Tooth loosening was associated with incident diabetes [adjusted relative risk = 1.73, 95% confidence interval = 1.14-2.64] after adjusting for other confounding factors. Gingival hemorrhage displayed a similar trend but was not significantly associated with incident diabetes [adjusted relative risk = 1.32, 95% confidence interval = 0.95-1.85]. CONCLUSIONS:Tooth loosening is an independent predictor of incident type 2 diabetes in Japanese men.