利用健保申報資料探討診斷和醫令的關係

Abstract

資料庫中同欄位(column)的資料(data)往往會隱含不同的屬性，而這些不同的屬性對應到其他特定欄位的資料也會隱藏一些非直接、不同意義上的關聯。越龐大的資料庫，各資料間隱含的資訊(information)越多且越複雜，想辦法從這些亂數中找出規則(rule)，這些規則可能就是屬性之間的關係，進而能從規則中找出例外或獨特性。使用系統化的流程和方法，只要使用者能夠先定義好要資料探勘的目標，即使並非該領域的專家也能分析出有意義和有趣的結果。

本論文的目標是在健保申報資料庫上，建立一個圖形化自動系統，利用資料探勘的概念，能夠從不同的病例中判斷各個醫令是歸屬該病例的那個診斷中。如此就能建立一個以醫令為基礎，評估各個診斷在不同病例的費用比率之介面。 系統除了能夠分辨診斷在醫令中的重要性以及診斷在病例中的重要性外，並建立了診斷醫令比較資料庫，能將所有醫令和所有診斷的關聯度做完整的呈現和查詢，提供一個住院醫令決策支援系統讓相關醫療工作者使用。

Each item among the same data column of database is often different from each other in the agent of hidden relation with other specific data column. The bigger database, the more data have implicit information and complex. To explore a way to find out the rules from these nonrandom relationship. These rules decide the relationship between attributes, and it can be need to find out the exceptions or define the uniqueness. By systematic approach and the usage of defined analysis methods, the user can first set the goal of data mining, then non-experts can analyze the database with meaningful and interesting result. The purpose of this paper is to use the National Health Insurance claim report database to build an automatic graphical system to elucidate the relationship between diagnosis and case management orders. We are able to establish an interface which can evaluate the contribution of case expense among their various diagnoses. This system can also identify the relative importance of each diagnosis in different order usage of any particular case. In addition, we have established an order and diagnosis relationship database which included all the possible correlation between diagnoses and orders. Which is so comprehensive and user friendly. It can provide an inpatient management decision support system for relevant health care workers.