CMStatistics 2017: Start Registration
View Submission - CMStatistics
B1202
Title: Calculating a generated effect modifier (GEM) for treatment selection based on imaging data Authors:  Todd Ogden - Columbia University (United States) [presenting]
Hyung Park - Columbia University (United States)
Eva Petkova - New York University (United States)
Thaddeus Tarpey - Wright State University (United States)
Abstract: A major goal in precision medicine is to make optimal patient-specific treatment decisions using data observed at baseline. For the treatment of neuropsychiatric disorders, available data may include clinical variables and measures of bevavioral/cognitive performance, as well as complex imaging data. We will present methods for (1) determining low-dimensional projections of all these data that are useful for describing differential treatment response; and (2) for estimating nonparametrically defined link functions based on these projections for each potential treatment. The resulting model can potentially provide powerful tools for precision medicine.