- 學位論文
應用資料探勘手法研究檢驗室失約原因
Using a Data Mining Approach to Identify Patient Classification According to No-Show Behaviors
摘要
近年來隨著醫療機構的整合與調整，為了配合政府在醫療政策上的改變，醫療產業不僅有外在的競爭環境，在醫院內部的控制與管理，都遇到了許多困難與挑戰，因此醫院的管理者渴望提升醫療資源使用上的效率，希望醫療資源與設備都能夠被妥善利用，減少醫療資源不必要的浪費。其研究目的是希望找出超音波檢驗室執行預約的病患，具有高風險失約率的病患特徵之外，並找出與出席與否有重要關聯的變數，並能夠針對執行超音波的病患進行分類，而在超音波檢驗室的失約率是否會因其病患特徵的不同，進而產生不同的失約率。主要研究方法以資料探勘分析執行超音波檢驗病患的失約情形，主要採用CHAID決策樹的分析方式，將執行超音波檢驗的病患進行分類動作，並找出執行超音波檢驗中失約病患的主要特徵因子，藉以了解在不同群組中的病患特徵。從研究結果顯示出，利用CHAID決策樹的分析方法，得知以病患的年齡用來判斷出席與否是最佳的病患特徵值，也找出其他與出席與否有重要關聯的變數為檢驗項目及性別因子，而利用CHAID決策樹的分析將研究對象分成五大群組，並產生五個主要分類法則。
並列摘要
Health care industries have become an increasingly critical sector for most developed countries, and Taiwan is no exception. After the implementation of National Health Insurance program, the tightly regulated health care funding approach requires hospitals and physicians to find effective ways to utilize resources. The competition between hospitals to attract patients has pushed them to provide better medical service with higher quality and more customer-oriented operation. In order to achieve better management performance while maintaining adequate quality of care, the coordination between medical professionals and the management team is the key. This proposal is to find patient classification based on their no-show rates. Using data mining technologies such as chi-square testing and chi-square automatic interaction detection (CHAID), factors associated with missed appointments are first identified and used to identify subgroups that are significantly different from another based on the factor values, or categories. The resulting tree from CHAID analysis includes 5 major categories, as age less than 35 being the category with this highest no-show rate. This analysis results could be useful for hospitalists to segment patients in terms of no-show probabilities for further improvement in utilization and quality of service.