Optimization-Based Data Mining

The goal of this project is to develop new optimization-based methods of data mining that can be applied to a wide range of problems of interest to us and to society. To evaluate our new techniques, we will apply them to optimizing drug combinations in the treatment of cancer. Our objective is to build models that utilize all available information related to all types of cancer, since many drugs are commonly used against multiple cancers. We intend to identify drug combinations that seem very promising and propose new clinical trials to test their effectiveness. In addition, we aspire to build a system that takes information about existing clinical trials and, together with historical information, decides adaptively how to combine trials, stop them early and start new ones. Early action is particularly important, as cancer patients do not have the luxury of time. Other potential application areas for this work include sensor management; intelligence, surveillance, and reconnaissance/strike operations; logistics; and traffic management.