A multi-site project is setting the foundation for a scalable, pan-Canadian synthetic data network (SDN) that allows hospitals to collaborate on research without sharing sensitive patient information.

The University of Waterloo is developing the SDN as part of a new cross-Canada consortium with institutions from Ontario, Quebec and New Brunswick. The SDN includes institutions such as the McGill University Health Centre, the University Health Network, the University of New Brunswick, the Canadian Personalized Healthcare Innovation Network, ZS Associates and MDClone Canada.

The University of Waterloo and the SDN are also collaborating with the Canadian Institute for Health Information (CIHI) and the National Research Council of Canada (NRC) on advancing synthetic data generation methods to support health system innovation while protecting privacy. 

The consortium was one of 14 groups nationally that received inaugural funding from the Digital Health Innovation Fund (DHIF), a new national program led by the Terry Fox Research Institute’s Digital Health and Discovery Platform. The DHIF program provides $900,000 in funding to the SDN partners, which will contribute $3.6 million to establish the network.

“Canada’s healthcare system produces an immense amount of data with the potential to drive better decisions and better outcomes,” said Helen Chen, Ph.D., a professor in the University of Waterloo’s School of Public Health Sciences, in a statement. “However, vital privacy requirements can make it difficult to share and use person-level data at scale. This limits collaboration and leaves important opportunities for discovery untapped.”

To address the issue, the consortium’s project will leverage synthetic data – computer-generated data that highly resemble statistical characteristics of the real patient data but contains no information that can be traced back to a specific person. Synthetic data is non-reversible and artificially created synthesis of real patient data. The dataset replicates the statistical characteristics and correlation of the raw data.

 “Synthetic data preserves the richness of variables while enhancing privacy protection,” Chen said. A federated learning approach will allow models to learn from synthetic data and maintain patient privacy.

The project will first focus on lung cancers using datasets from the McGill University Health Centre and the Princess Margaret University Health Network. The team will use the datasets to better understand patient populations, care pathways, treatment patterns and healthcare resource use. They will then test whether the approach can be scaled up for broader use.