Abstract:
|
With new treatments and advanced genomic technology available, personalized medicine has become an important piece in the new era of medical product development. When the treatment effect varies across sub-populations, it would be desirable to develop individualized treatment rules (ITR) according to individual patients' baseline characteristics. So far building ITR is mostly based on the information from randomized clinical trials (RCT). However based on exclusion and inclusion rules, RCT data may not reflect real world treatment effectiveness. In this paper, we develop ITR using healthcare data with patients' genomic profiles based on a new optimization scheme. As the large scale and high dimensional features of healthcare and genomic information lay out serious computational burden, we further solve the problem by an efficient tree algorithm. Our method is a simple decision that clinician can easily adopt. ITR performances are evaluated in end in terms of how to better calibrate genomic information. Data applications to veteran affair healthcare data demonstrate utility of our method.
|