TrialProbe is a toolkit for probing non-experimental methods for inferring cauasal effects using ClinicalTrials.gov data.
This codebase consists of two components which correspond to two folders, extraction and analysis.
create_reference_set is Python and Julia code for extracting, cleaning, and preprocessing ClinicalTrials.gov data into a simple JSON format, reference_set.txt
To encourage ease of use, an example reference_set.txt from that pipeline can be found in this folder.
analyze_methods is Python code for analyzing non-experimental methods using reference_set.txt from the create_refererence_set pipeline.
The recommended entry point for new users is analyze_methods/plot.ipynb, as it illustrates how to use this reference set to compute concordance.