This week, a team at Rambam Health Care Campus published a paper in JMIR Medical Informatics describing the value of MDClone and synthetic data to democratize data, providing the ability to conduct “rapid, safe, and repeatable analysis of data in hospitals.”
The paper shares results from five studies at Rambam where researchers used MDClone to build patient cohorts and extract synthetic data for analysis before returning to MDClone to extract original data for validation of the results. In each case, the synthetic data “provides a close estimate to real data results.”
The data access process described in the paper is transformative for Rambam, the first health system to implement MDClone. It provides a self-service model for researchers to quickly test hypotheses using synthetic data and to use PHI (with an IRB) only when needed, prior to publication or the treatment of patients. As the research shows, because the synthetic data closely resembles PHI, in many cases − including in quality analyses or other hospital operations-oriented initiatives − there is no need to ever use PHI.
The analysis included logistic stepwise regression with 11 features and all their 2nd order interactions, Survival analysis using both Kaplan-Meier and a multivariate Cox proportional hazard regression, propensity score matching, and more. They were applied to the study of Protein Pump Inhibitors Prescription, Percutaneous Coronary Intervention and ST-Elevation Myocardial Infarction, Blood Urea Nitrogen and Acute Decompensated Heart Failure, Imaging Nephropathy, and Hypoglycemia Insulin.
“As we expand usage of MDClone with our current customers and launch MDClone in more sites, this paper, and Rambam’s ongoing use of MDClone, will provide valuable insight for organizations who want to leverage their own data assets more efficiently and safely than ever before,” says Hovav Dror, PhD, Former Chief Science Officer of MDClone.
Read the full research paper here: https://medinform.jmir.org/2020/2/e16492/
