If we want to achieve quality, efficiency, and transparency of observational research, we need to standardize the structure, content, and analytics in such a way that they can optimally support our specific use cases. We have adopted the OMOP common data model (CDM) as a platform for doing that. We are also developing special open-source tools and processes to help you implement the OMOP CDM and the associated standardized analytics tools within your institutions.
Medical product safety surveillance
Mankind ought to have a way to know and evaluate potential associations between the exposure to any medical product and all its adverse effects. We want to address this gap by establishing an international risk identification and analysis system available to the public. It will enable proactive detection of potential drug effects. To that end, we are currently developing large-scale analytics that allow for the real-time exploration of all medical products and all health outcomes of interest. We are also establishing an open-source evidence repository so that anybody or any organization with observational data who wishes to contribute to that objective can produce (using our open-source tools) and share their evidence, so we can all learn from each other.
Comparative effectiveness research
Similarly, the public also deserves to understand alternative treatments and to be able to make direct comparisons between the effectiveness of those alternatives. We are developing open-source tools to generate this evidence from observational health data. Unlike our safety surveillance tools, which seek to address the question, “Does this medical product cause that outcome?”, our CER tools seek to answer, “Does this medical product cause the outcome more or less than alternative products I could consider for the same purpose?”.
Personalized risk prediction
Patient-level predictive modeling can complement population-level estimation, to go beyond answering societal and policy questions about average treatment effects to deliver individualized insights about your personalized future risk of experiencing an outcome given what we know about you in the past: your specific demographics, your medical history and your prior health behaviors. We are developing advanced algorithms to be able to produce these personalized estimates, and are delivering tools to communicate this risk information directly to patients.
To learn from observational health data and generate reliable and transparent real-world evidence, we must understand the underlying source data. Much of our research involves analysis of data that were captured for purposes other than research – insurance claims are financial transaction data to support the provider-payer reimbursement process, electronic health records are systems to support providers in their clinical care. Therefore, it is critically important to understand where the data come from and why, to help interpret what it all means. We are developing tools for data quality assessment and database profiling so that you can know when a database can and cannot be used, and what data issues require consideration in any analysis.
Health systems continuously seek to improve the quality of care provided to their patients. We are developing open-source tools to make that job easier through the systematic application of quality measures within observational data in the OMOP common data model. We are supporting these efforts through the design, development, and evaluation of new empirical-evidence-based quality measures.