Locally Smoothed Statistical Learning for Age-Dependent Classification and Disease Risk Prediction

• Activity Number: 393
• Type: Invited
• Date/Time: Tuesday, August 6, 2013 : 2:00 PM to 3:50 PM
• Sponsor: ENAR
• Author(s): Huaihou Chen and Tianle Chen and Donglin Zeng and Yuanjia Wang*+
• Companies: New York University and Columbia University and The University of North Carolina and Columbia University
• Keywords: Biomarker studies ; Large margin based classification ; High-dimensional data ; Kernel smoothing ; Risk bound

Abstract

Accurately predicting whether a presymptomatic individual is at risk of a disease based on marker profiles offers an opportunity for early intervention well before diagnosis. For many diseases, the risk of disease varies with age and with markers that themselves may be age dependent. Markers predictive of a younger subject's risk status may be different from markers predictive of an older subject. Therefore, developing age-dependent prediction rules is valuable in improving prediction accuracy and guiding timely intervention. To identify effective classification and prediction rules using nonparametric decision functions, standard learning approach that includes age interchangeably with other markers as input variables may be inadequate in singling out the age effect, especially for high-dimensional markers. In this work, we propose a local smoothing large margin classifier implemented with support vector machine (SVM) to construct effective age-dependent classification rules. The method adaptively adjusts age effect and treats age and other markers separately. We derive the asymptotic risk bound of the local smoothing SVM, and perform extensive simulation studies in order to comp