a sequential logistic regression classifier based on mixed effects with applications to longitudinal data：基于混合效应的时序逻辑回归分类器在纵向数据中的应用.pdfVIP

Computational Statistics and Data Analysis 94 (2016) 238–249 Contents lists available at ScienceDirect Computational Statistics and Data Analysis journal homepage: /locate/csda A sequential logistic regression classifier based on mixed effects with applications to longitudinal data Xin Zhang a , Daniel R. Jeske a,∗, Jun Li a , Vance Wong b a Department of Statistics, University of California—Riverside, Riverside, CA 92521, USA b Alere Inc., San Diego, CA 92121, USA a r t i c l e i n f o a b s t r a c t Article history: Making an early classification in longitudinal data is highly desirable. For this purpose, a Received 16 May 2014 sequential classifier that incorporates a neutral zone framework is proposed. The classi- Received in revised form 18 August 2015 fication procedure evaluates each subject sequentially at each longitudinal time point. If Accepted 18 August 2015 there is not adequate confidence in making a classification at a given time point, the de- Available online 28 August 2015 cision will wait until the next time point where another measurement is collected. This process continues until there is enough confidence of making a classification or until the Keywords: Classification last time point where data can be collected is reached. It is demonstrated that the proposed Mixed effects models sequential classifier maintains competitive error rates while reducing the overall cost when Longitudinal data the cost of ti