頁籤選單縮合
題名 | 風險分類與健保個人醫療費用預測=Risk Classification and Prediction of Individual Health Expenses |
---|---|
作 者 | 許碩芬; 林兆欣; 楊雅玲; | 書刊名 | 管理評論 |
卷期 | 25:4 民95.10 |
頁次 | 頁27-48 |
分類號 | 419.45 |
關鍵詞 | 風險分類; 自組織映射網路; 倒傳遞網路; 醫療費用預測; Risk classification; SOM; BPN; Prediction of medical expenses; |
語文 | 中文(Chinese) |
中文摘要 | 台灣健保已經全面實施總額預算支付制度,以有效抑制醫療費用的成長,且在未來有朝向多元健保人體制之規劃。為了改善總額預算額度分配及競爭型健保市場的公平性和效率性,本文結合人工智慧方法中的自組織映射網路(Self-Organizing Map network, SOM)和倒傳遞網路(Back-Propagation Network, BPN)以預測個人醫療費用,作為健保預算分配依據,並作為未來引進多元健保人制度時,計算風險調整保費的基礎。 本文除利用上述研究方法來預測個人醫療費用外,並比較各種模型之預測能力,以驗證此研究方法的可行性。結果發現:(1)鑑於健保資料分配的特殊性,進行風險分類可增加模型之預測能力。(2)全民健保保險對象之風險分類可以SOM分為兩類。(3)結合SOM和BPN比單獨使用BPN或線性迴歸模型之預測能力高。 |
英文摘要 | In order to solve the trade-off relationship between the Law of Large Numbers and homogeneity principal in risk classification, we proposed a methodology by combing two neural network model s, i .e., Self-Organizing Map network (SOM) and Back-Propagation Network (BPN). This paper uses empirical data from Taiwan's Nat ional Health Insurance (NHI) to demonstrate that the healthcare expenses budget assignment under global budget system and the cream-skimming problem under multi-carrier health market can be solved and hence fairness and efficiency can be improved by combining SOM and BPN. In addition, this paper examines the predictive power of different medical expenses predict ion models. Main results obtained from this paper are as follows: (1) risk classification can improve the predictive power of forecasting model. (2) Individuals of NHI can be classified into two subgroups by SOM. (3)The predictive power of combination model of SOM and BPN is better than using BPN alone or linear regression alone. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。