Title: Semiparametric mixed model for analysis of covariance with high dimensional covariates
Authors: Erniel Barrios - University of the Philippines (Philippines) [presenting]
Stephen Jun Villejo - University of the Philippines (Philippines)
Joseph Ryan Lansangan - University of the Philippines (Philippines)
Abstract: Treatment effects are difficult to measure from clinical trials involving live human subjects due to the contamination of the response resulting that non-homogeneous experimental units. The experiment should be designed to allow measurement of as many covariates as possible, so that responses are adjusted to facilitate estimation of treatment effects. There could be more covariates than the number of experimental units that manifest the responses, leading to a potentially overparameterized model and prone to observing false positive evidence on treatment effect. We consider an additive mixed semiparametric model to adjust the responses due to heterogeneity of experimental units indexed by the covariates. We then simultaneously estimate the treatment effect while simultaneously reducing the dimension of the covariates. Simulation studies are conducted, the methods are also used in actual clinical trials.