Special Scientific Procedures: Statistical Modeling Core

招募和收集大量学科的多学科数据的努力是没有意义的

The effort to recruit and collect multidisciplinary data on a large number of subjects is of most benefit when attempting to use this information to provide practical guidelines for clinicians treating patients. In order to accomplish this, we will need to test the possible interactions among separate measures such as brain function, stress reactivity and underlying genetic structure. The purpose of this core will be to use a combination of model building techniques, similar to those used in models of medical risk, and a comprehensive collection of relevant biological and clinical measurements to begin the process of developing reliable models of treatment response for depressed patients. We will use state-of-the-art variable selection techniques to build and validate a clinical model for overall and treatment-specific response, and evaluate its predictive ability through the use of statistical techniques. In addition, we will be incorporating genetic information, which will require specialized estimation techniques in order to model the relationship with treatment response. Thus we will have a statistical geneticist (Michael Epstein, PhD) working closely with the core to provide expertise and supervision for any treatment response related analyses, and a specific member of the core designated for handling those techniques.