Characteristics of Type-2 Diabetics Who are Prone to High-Cost Medical Care Expenses by Bayesian Network

Abstract

Objective: This study aims to determine the characteristics of Type 2 diabetic patients who are more likely to cause high-cost medical expenses using the Bayesian network model. Methods: The 2011–2015 receipt data of Iwamizawa city, Japan were collected from the National Health Insurance Database. From the record, we identified patients with Type 2 diabetes with the following items: age, gender, area, number of days provided medical services, number of diseases, number of medical examinations, annual healthcare expenditures, and the presence or absence of hospitalization. The Bayesian network model was applied to identify the characteristics of the patients, and four observed values were changed using a model for patients who paid at least 3607 USD a year for medical expenses. The changes in the conditional probability of the annual healthcare expenditures and changes in the percentage of patients with high-cost medical expenses were analyzed. Results: After changing the observed value, the percentage of patients with high-cost medical expense reimbursement increased when the following four conditions were applied: the patient “has ever been hospitalized”, “had been provided medical services at least 18 days a year”, “had at least 14 diseases listed on medical insurance receipts”, and “has not had specific health checkups in five years”. Conclusions: To prevent an excessive rise in healthcare expenditures in Type 2 diabetic patients, measures against complications and promoting encouragement for them to undergo specific health checkups are considered as effective.