Friday, February 19, 2016


Topic: Predictive methodology mobForest (use in personalized treatment)


Presenter: Georgiy Bobashev


Recently we have developed and validated methodology model-based Forest (mobForest) that allows one to incorporate “What if” scenarios into tree-based forest models. Such questions arise when it is important to predict potential patient’s outcome if the patient is to be assigned to an alternative treatment. I will illustrate the use of this methodology From a largest alcohol treatment clinical trial called COMBINE. Our methodology allows one to forecast patient’s outcomes from alternative treatments and evaluate the proportion of individuals who would benefit from personalized treatment and identify characteristics of such population.