Purpose
Health data holds the promise of improved treatments and better patient outcomes, but is fragmented, not interoperable and inaccessible. If we could transform our healthcare system to enable meaningful data translation at scale, the system would benefit significantly by being able to act on insights, ultimately benefiting patients. For example, as over 90% of patients are not eligible for clinical trials there is a growing recognition that real-world data (RWD) should be used to replicate and extend trial results. Despite enactment of the 21st Century Cures Act requiring the FDA to release guidelines on how RWD can inform regulatory decisions, accessing this data and ensuring it is of high quality remains a significant challenge. There are currently no widely accepted alternatives to manual chart review for abstracting RWD from free-text documentation and few stakeholders have the budgets required to pay for it. Furthermore, the process is time consuming, inconsistent and produces siloed data repositories that are seldom linked to provide pan-Canadian RWD that could be used to transform the healthcare system. This new initiative endeavours to address these challenges and transform the healthcare system through data, enabling better care and health outcomes for patients from coast to coast to coast.
Approach
Public and private healthcare stakeholders have been convened to orchestrate initiatives utilizing novel Artificial Intelligence (AI) and Natural Language Processing (NLP) to abstract and analyze real-world data from clinician documentation. These partners, will form a pan-Canadian integrated data network which will serve as a digital backbone to collect, connect and make available anonymized data to accelerate system transformation. This new initiative will accomplish this is through Programs demonstrating the value of RWD integration, harmonization, portability, and accessibility for use by clinics, researchers, and innovators.
Findings
The first of two initial Programs will determine the sequencing of testing and treatment that produces the best health outcomes for advanced lung cancer patients. Data from a single site will be analyzed and compared to data from across Alberta and then other provinces, to determine intra- and inter-provincial variations in care and whether variations are securing better outcomes for patients. The pilot for this program successfully abstracted and analyzed RWD on over 1200 patients from a single site in Ontario and the work was presented at the World Conference on Lung Cancer. The program is expecting results in January 2020 and we are excited to share these with e-Heath 2020.
Conclusion
We describe the challenges and a novel application of the latest AI and NLP techniques to abstract high-quality RWD from clinician documentation at a significantly greater scale and lower cost to current manual chart processes. Dramatically reducing the cost to abstract and analyze RWD will enable this pan-Canadian multi-stakeholder initiative to leverage RWD to make a positive difference in the lives of patients. The benefits will be nationwide: we will analyze data housed at institutions across Canada and make the insights available across Canada. Transparency will lead to equitability in knowledge, insight, access, and most importantly, benefits for patients.
