運用資料探勘分析門診病歷資料－以某區域醫院小兒科為例

本研究利用小兒科門診資料，分為三大主題方向進行研究探討。（1）由於每位醫師的訓練背景不同，對於相同藥品使用頻率的認知也略有差異，因此藥品的使用頻率是否會影響病患回診時間長短，成為本研究探討主題之一。（2）探討政府醫療補助計畫與病患就醫行為之關係─部分負擔費用提高後，是否影響病患醫療層級的選擇。（3）門診兒科醫師與急診兒科醫師診斷之差異性比較，以及兩者對於醫療資源的使用情形。根據研究結果顯示：（1）利用資料探勘（Data Mining）分析出來的結果發現，藥品使用頻率與病患的回診狀況並沒有絕對性的關係，也許是個人體質、外在環境條件、亦或其它疾病併發症影響導致。（2）醫療費用高低不一定是病患就醫最重要的考量因素，對於醫師的信賴、距離的遠近等，都是影響條件。（3）在2001年7月之後，病患支付檢驗檢查費呈現遞增趨勢，顯示急診兒科醫師使用檢驗檢查項目輔助診斷的比例較高。本研究為了解醫療資源的使用情形，透過資料探勘與統計方法針對三大主題加以分析，並將其結果提供給醫院、醫療相關人員做為參考依據。

關鍵字

資料探勘；用藥頻率；門診兒科

並列摘要

This research made use of the outpatient data of pediatrics, and was divided into three main subjects to debate. （1） Because of different training background for each doctor, so there were some cognitive differences to using frequency of the same medicine, therefore, will the frequency of the patient's medicine usage affect the rate of their future revisit examination? （2） To discuss the relationship between medical allowances of patients and their actions of see a doctor─will patient change their choice of medical treatment level, that when medical copayment be raise up? （3） Comparing the diagnostic differences between clinic pediatrician and emergency pediatrician, as well as their patterns of usage in medical resources. The results of research show that: （1） The result of Data-Mining found that; the frequency of the patient's medicine usage has no absolute relationship with their situation of revisit, perhaps, this is relating to personal physique, conditions of external environment, or complications caused to other disease. （2） When patient go to a doctor, the expense is not necessarily the most important consideration, the confidence to doctor, or far and near distance…etc all are considerations. （3） After July 2001, patients paid the expense for tested-check items was displayed raise up, it is show that emergency pediatrician use tested-check items to assist diagnosis have higher proportion than clinic pediatrician. In order to understand the situation of usage in medical resources, this study was directed to three main subjects, and analyzed by Data-Mining and statistical methods, and provide the result to hospital and medical-related staffs as a reference.