In this talk, Gil will share some of the principles and techniques used by Flatiron to build the nation’s largest cancer database. Flatiron Health’s platform connects cancer centers across the country on a common technology infrastructure, supporting treatment for about 1 in 5 active cancer patients in the U.S. Their architecture is focused on enabling rapid development iterations with cancer specialists, nurses, and other stakeholders. Rapidly integrating medical data from hundreds of heterogeneous source systems required Flatiron to develop a unique data integration and processing framework. Iterating quickly while incorporating knowledge from oncologists, nurses, medical informaticists, and healthcare analysts has largely shaped this architecture as well.

The talk will focus on tooling and the architecture that we built to support rapid iteration on complex ETL processes (iteration itself is quick) and minimizes knowledge translation iterations (number of iterations is smaller - we get stuff right with less iterations). The lessons taught here are on building architecture that optimally supports a business process or a unique development process and less about CS concepts like scalability, reliability etc.