Development of Health Insurance Rate-Setting System Utilizing Data Mining Technique

Yong, Wang Sik; Chae, Young Moon; Kang, Sung Hong; Kim, Won Joong

Abstract

The objective of this study is to present a new approach to the development of national health insurance rate-setting system using data mining technique. In particular, this paper examines the feasibility of applying the rate-setting system for the regional insurance program to the insureds of the workman' s insurance program, so that the unified rate-setting system can be developed. The subjects of the study were the beneficiaries(sample size = 234,224) of the workman' s insurance program(excluding government officials and teachers) residing in Goyang City. 1998 data from Korean Medical Insurance Corporation were analyzed through various data mining techniques such as Decision Tree, Regression, Assessment and SCORE. Major results and conclusions are as follows: 1) On average, the insured' s property element of insurance fee, automobile element of insurance fee, property score of income assessment, and automobile score of income assessment together accounted for only 16% of the actual income, which implies that we need more precise data in order to develop an equitable rate-setting system. 2) To do so, it is necessary to gather all the income-related data(property income, financial income, earnings, ...) in a timely manner. Since current computer and information system would not allow efficient maintenance and utilization of such data, it would also be necessary to establish data warehouse for this purpose. 3) In developing the rate-setting system on the basis of data warehouse, various data mining techniques should be employed to ensure the equity of insurance rates among different classes of insureds.