課程一:Parallel Groups Design versus Crossover Design: Definitions, Ideas and Limitations。課程二:Statistical methods in the study on suboptimal health state with Traditional Chinese Medicine
開始日期
2015-05-14 14:00:00
結束日期
2015-05-14 17:00:00
活動說明
【課程一】
主題：
Parallel Groups Design versus Crossover Design: Definitions, Ideas and Limitations
課程介紹：
Parallel Groups Design and Crossover Design are two most commonly-employed designs in randomized clinical trials. The general ideas and principles behind the parallel groups and crossover designs are presented. Simple statistical models are used to provide audience with an insight into when and how using the crossover design can gain efficiency as compared with the parallel groups design. Advantages and disadvantages between using these two designs from the practical point of view are also given.
講師簡介：呂剛中教授 (Fellow of ASA & ACE)
San Diego State University, USA
時 間：104年5月14日(四) 14:00-15:30
地 點：醫綜前棟三樓會議室
【課程二】
主題：
Statistical methods in the study on suboptimal health state with Traditional Chinese Medicine
課程介紹：
Suboptimal health state is a public health problem all over the world. According to the recent investigation by the World Health Organization, more than 70% of the world population suffers from the suboptimal health problem. Compared with the Western Medicine, the Traditional Chinese Medicine has its unique advantages on the treatment to the suboptimal health state because the TCM doctor possesses many theories and methods for diagnosing, preventing and treating suboptimal health conditions. But the imbalanced data structure always leads to the problems of low true positive rate and inconsistent significant variables for the commonly used statistical classifiers. In this talk, we propose two methods for both the classification and feature selection studies with imbalanced data problem. Numerical study shows that the two proposed methods can significantly enhance the classification accuracy for the minority class with small loss on the true negative rate on the majority class while generating high true positive rate and select truly significant variables simultaneously for both continuous and categorical covariates. Illustrative example on the analysis of the suboptimal health state data in the traditional Chinese medicine is discussed to show the reasonable application of the proposed methods.
講師簡介：李揚副教授
中國人民大學
時 間：104年5月14日(四) 15:30-17:00
地 點：醫綜前棟三樓會議室
【對象】
全校教職員、三院同仁及學生
【承辦單位】
生物統計研究中心
【聯絡窗口】
吳佩真(ext8670)
