Quantifying representativeness in RCTs using ML fairness metrics - Data and codes

The "Quantifying representativeness in RCTs using ML
fairness metrics - Data and codes" is used to quantify
representativeness in randomized clinical trials (RCTs) and provide
insights to improve the clinical trial equity and health
equity. We developed RCT representativeness metrics based on
Machine Learning (ML) Fairness Research. Visualizations and statistical
tests based on proposed metrics enable researchers and physicians to
rapidly visualize and assess subgroup representation in RCTs. The approach
enables users to determine underrepresentation, absence, or other
misrepresentation of subgroups indicating potential limitations of RCTs.
The method could help support generalizability evaluation of existing RCT
cohorts, enrollment target decisions for new RCTs (if eligibility criteria
are included), and monitoring of RCT enrollment, ultimately contributing
to more equitable public health outcomes. We apply the proposed
RCT representativeness metrics to three landmark clinical trials released
in the last decade: Action to Control Cardiovascular Risk in Diabetes
(ACCOD), Antihypertensive and Lipid-Lowering Treatment to Prevent Heart
Attack Trial (ALLHAT), and Systolic Blood Pressure Intervention Trial
(SPRINT). This dataset contains the processed data and results for the
experiments and visualization codes in the paper titled "Quantifying
representativeness in randomized clinical trials using machine learning
fairness metrics."