A0966
Title: Joint semiparametric kernel machine network regression
Authors: Byung-Jun Kim - Michigan Technological University (United States) [presenting]
Inyoung Kim - Virginia Tech (United States)
Abstract: A joint semiparametric kernel network regression model is developed for possibly nonlinear or non-additive associations and complicated interactions on both variable selection and network estimation. The approach is a unified and integrated method that can simultaneously identify important variables for a continuous outcome and build a network among the variables. The advantages of our proposed approach lie in flexibility, interactivity, and interpretability through the connection between the variable selection and the network estimation. We demonstrate our approach using real data application on genetic pathway-based analysis.
