Clinical Dataset Causal Inference Benchmarks- Summer 2016


Clinical Dataset Causal Inference Benchmarks

Supervisor: Dr. Peter Szolovits, CSAIL MEDG

Contact: Marzyeh Ghassemi,



In this project, we will create a benchmark for state of the art causal inference methods. Datasets will be extracted from the public MIMICIII clinical database, and cohorts will be based on published papers that specify randomized control trials (RCTs). This is an opportunity to create electronic RCTs to explore meaningful clinical questions that often cannot be answered with a traditional RCT. 



Student should be comfortable developing in SQL or using Python to extract datasets, and capable of contributing to a Github repo.