Assistant Investigator Radiation Oncology, Mass General Research Institute

algorithms; biological network modeling; convex optimization; genomics; kernalized support vector machines; machine learning; mathematics; mixed integer programming; nonlinear approaches; radiation for cancer control; radiometry; radiotherapy conformal; radiotherapy dosage; radiotherapy intensity-modulated; radiotherapy planning computer-assisted; random forest; statistics; systems biology

My current focus is on the development of machine learning approaches for personalized cancer treatment. The main research questions are: what is the value of prior knowledge in this domain and how can we best incorporate such expert knowledge. We have recently published two papers that clearly demonstrate that for complex systems it is highly useful to incorporate expert knowledge when using machine learning to predict the system behaviors (see here [supplementary info] and here).

We are also pursuing cell line drug and radiation sensitivity datasets as a critical step towards the eventual development of a clinical personalized drug selection tool for cancer. This research is ongoing but please reach out to me at [reversed] broadinstitute.org [at] dcraft to discuss directions, pitfalls, new ideas, etc. 

In the world of radiation therapy, I consider VMAT still an unsolved problem. It is highly non-convex, and yet there are many convex aspects to the problem. My group is working to tame those non convexities by smart heuristics.