Sparse Partial Functional Linear Regression Model for Hyper-Acute Ischemic Stroke Study

• Activity Number: 226
• Type: Topic Contributed
• Date/Time: Monday, August 4, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Physical and Engineering Sciences
• Author(s): Linglong Kong*+ and Hongtu Zhu and Hongyu An and Andria Ford
• Companies: University of Alberta and University of North Carolina at Chapel Hill and University of North Carolina and Washington University in St. Louis
• Keywords: Partial Functional Linear Regression ; Hyper-Acute Ischemic Stroke ; Sparse estimation ; B-Splines

Abstract

In Hyper-acute ischemic stroke study, it is critical to assess tissue perfusion during hyper acute stroke using certain MRI parameters. It has been shown in various literatures that mean transit time (MTT) in general is a better measure than time-to-peak (TTP), and time-to-maximum (Tmax) to best predict neurological improvement and tissue salvage following early reperfusion. However, at what time point or interval should the MTT be measured remains unclear. In this talk, we propose to use group adaptive LASSO to simultaneously estimate the sparse functional coefficients of MTT at different time to find the best measure time interval. We use a perturbation bootstrap method to obtain the $p$ values for each function basis. Simulation studies show that our proposed method outperforms others. We apply it to a real data analysis on Hyper-acute ischemic stroke study. The results provide some new insights in addition to confirming some previous findings.