醫學系學生第一階段醫師國家考試與學習表現之相關性探討－以北部某國立醫學院為例

The Correlation between Medical Students' Learning Performance to Taiwan Medical Licensing Examination Step1- Example by the National Medical College in North of Taiwan

閔善柔(S.R. Min) ； 朱基銘(C.M. Chu)

《醫療資訊雜誌》 26卷3期 (2017/09) Pp. 15-24

摘要

第一階段醫師國家考試是醫學系學生為考取醫師執照的第一道門檻，現今不論在美國或是台灣，皆非常重視醫學系學生在校期間學習表現是否符合未來臨床知能需求，並因應隨著醫學新知的日益月新而精進之教學方式，力倡於學生在校期間能及早偵測到需要介入輔導的學生策略，以提高醫學系學生於醫師國家考試的通過率。故如何運用已知的預測模型及早偵測學生考前需介入輔導的最佳時機，有效降低醫學系學生未通過醫師國家考試的風險，仍有探討空間。本研究旨在利用巨量資料的概念，採回溯性分析(Retrospective analysis)，回溯自2003年至2016年蒐集北部某國立醫學院共690位醫學生的在校學習成績、數位學習教材下載使用情形、第一階段醫師國家考試結果等，串聯整併成一次級學務資料庫，並利用一系列嚴謹的文獻探討，研擬研究流程、設計、方法及數據分析後與其他已發表的相關研究結果比對，針對醫學系學生於西醫第一階段醫師國家考試前兩年－在校三、四年級時的學習表現與當年初次第一階段醫師國家考試通過與否結果之間的相關性探討以及進一步以ROC曲線建立最佳切截分數（cutoff scores）作為成績閾值（Threshold），預測警示成績閾值並以卜瓦松迴歸評估其迴歸模型解釋力。研究結果顯示，三、四年級在校學期總平均成績、考前的學習表現與第一階段醫師國家考試通過與否具相關性，第一階段醫師國家考試考試科目相關之各基礎醫學學科成績閾值每增加1分則通過第一階段醫師國家考試的預測機率之模型解釋力提高1.01~1.05倍(95%CI, 1.00~1.09)。醫學教學單位可設定各科成績閾值，將低於閾值的醫學生列為考前輔導對象。

關鍵字

醫學教育 ； 學習表現 ； 醫師第一階段醫師國家考試 ； 巨量資料 ； ROC曲線分析 ； 卜瓦松迴歸

並列摘要

The Medical Licensing Examination Step 1 exams the first threshold for medical students to obtain a physician's license, no matter if they are in the United States or Taiwan, that all the accent is on the learning performance of medical students for fit the requirements of clinical skill and sense, but not much researches has been published in Taiwan, there is a lot of possibly about how to use the known prediction model to instead of detect the students before the exam to be the best time to intervene in counseling, effectively reduce the failure risk of medical students on Taiwan Medicine Licensing Examination Step 1(TWMLE Step 1). The aim of this study is to determine whether correlation exists between the learning performances of medical students and Taiwan Medicine Licensing Examination Step 1 (TWMLE Step 1). And further explore the feasibility of establishing the best cutoff score as a ＂Threshold＂ of learning performance in basic medical compulsory subjects with ROC (Receiver Operating Characteristic, ROC) curve analysis and determining its accuracy as an accessible tool. In Basic medical compulsory subjects. This retrospective study enrolled 690 medical students of a National medical college in northern of Taiwan since 2003 to 2016. Cohorts were 690 third-years to fourth-years medical students over a 2 years period in scores of represented as performance from the secondary student status records to establishment of a complete secondary database, then were concept of the Big data is not absolute, but rather the relative and analyzed to determine the correlation with Taiwan Medicine Licensing Examination Step 1 (passage or failure) of the first year. The results showed that only the semester average score during third and fourth grade was highly significantly correlated with Taiwan Medicine Licensing Examination Step 1 (p＜0.001). A ROC (Receiver Operating Characteristic, ROC) Curve analysis and Poisson Regression where be used of verify the hypothesis of predict performance with cutoff score to be the threshold of Taiwan Medicine Licensing Examination Step 1 in the first year on research. For every extra of per cutoff score on research, 1.01~1.05 (95% CI, 1.00 to 1.09) times stronger predictor performance of Taiwan Medicine Licensing Examination Step 1, a statistically significant result, p＜0.001. In this study result revealed a positive correlation exits between the performance of scores during the third and fourth grade, and the ROC curve analysis was verified by the regression model. Accuracy of the method can provide medical teaching units, can also be counseling program for the medical students apply for the Taiwan Medicine Licensing Examination Step 1 in fist year.

並列關鍵字

Medical education ； Learning performance ； Taiwan Medicine Licensing Step 1 (TWMLE Step 1) ； Big Data ； ROC curve analysis ； Poisson Regression