Analyzing Observational Healthcare Data
Real World Evidence (RWE) refers to the data gathered on patients through the practice of medicine and patient care, from electronic health records, claim records and more. As electronic health records and health insurance claims of millions of patients become more available for population health studies, they offer incredible opportunities for Health Economics and Outcome Research (HEOR) and personalized treatment recommendation systems.
In our research, we use and develop machine learning and causal inference to analyze large observational healthcare data for various use cases, including decision support, drug efficacy, and drug repurposing in several disease areas. In conjunction with our RWE research project, we built a unique stack of analytics tools to support semi-automatic data mining and analytics workflows for RWE analysis.
As part of this work we have developed an open-source Python package that allows training various causal inference methods on RWE. The package allows choosing the underlying machine learning model that is used in the causal inference method, and provides a set of evaluation metrics and plots to estimate the performance and robustness of the method.
Some of our published work include RWE analysis in domains like epilepsy, diabetes type 2, mental illnesses, HIV, oncology, and more. For a complete set of published work and press releases see our publications page.