May 2015 Rimma Pivovarov This repository contains code created for Phenome model version 1. More on the phenome model in: Pivovarov R, Perotte AJ, Grave E, Angiolillo J, Wiggins CH, Elhadad N. Learning Probabilistic Phenotypes from Heterogeneous EHR Data. The input to the inference engine is a few different files that can be created from any EHR dataset. The input files are in the "data" directory, the "examples" directory there is just a set of example data with which to run the inference engine. Please note: the examples are completely made up and will not produce reasonable phenotypes. The results directory will collect all of the results of the learned phenome model. This includes serialized variables saved every few iterations and every time a maximum log likelihood is reached. In addition, the log likelihood numbers are saved along with the learned phenotypes at each maximum log likelihood.