topic model in emr project

I did this project during my summer intern in 2017 at Duke. It was the first time I got the knowledge of Electronic Medical Record and ICD-9 codes. We focused on Duke Medical Center Electronic Medical Records from 2004 to 2013 of 210,329 patients with 10,804 unique ICD9 diagnosis codes records for each patient. The algorithm I used is the supervised Latent Dirichlet Allocation(supvised topic model). The attached is the poster of this project.