Avillach, Murphy, et al., discuss the challenge in going from the dimensional tables of i2b2 (and similar data stores) to the sparse table representation that is expected in most instantiations of the PIC-SURE API.
Karen Copenhaver from Linux Foundation and Choate gave us the big picture (pun intended) on open source software, open source ecosystem and the different licenses. We then focused on the license characteristics we need for the PIC-SURE system and how to select it.
Griffin Weber described his experiments on a 2 million patient database spread across different hospitals to test different patient-joining/matching algorithms. Using dates vs concepts vs number of facts and the various combinations thereof have interestingly different performance characteristics.