Search

The objective of this project is to implement a harmonized artificial intelligence (AI)-based de-identification of free-text medical data across multiple Canadian jurisdictions. This federated learning approach will allow these jurisdictions to leverage each other’s data and resources while no individual-level data leaves the jurisdiction. Federated Learning enables health data centers in different jurisdictions to collaborate in training machine learning models without sharing individual-level data. This approach will significantly reduce privacy and cybersecurity risks and barriers that are involved in sharing and moving data across different jurisdictions. In a federated learning environment, machine learning models are trained on multiple data sources available in local data centers; local data are not shared to a central computing/analysis environment. Instead, parameters (such as model weights) are shared between these local data centers to generate a global model that will be shared and used by all participating data centers. In this case study, four health research data centers in different Canadian provinces will take part in deployment of an AI-based application for de-identification of free-text data. The data centers are members of Health Data Research Network (HDRN) Canada. The deployment will include: • harmonized annotation and labeling of local data, • local training of entity recognition algorithms, • integrating model weights from each data centers to create a global model • development of license agreements between the participating data centers to allow sharing model weights This is an ongoing project. The talk will demonstrate learning experiences, advantages, and challenges in a federated learning environment and explore the feasibility of transporting this approach to other multi-jurisdiction research networks.

Objectives People with disability represent 1 in 6 Australians and use services provided across multiple levels of government. Data is siloed between different governments and service systems, limiting visibility and ability to improve outcomes. Our objective was to progress people-centred data sharing across jurisdictions at scale, focused on people with disability. Approach More than 140 engagements with the disability community, governments, researchers, and service providers were conducted to ascertain user and data needs; systematically map enablers and barriers to data use; and outline the requisite technical capabilities and social license for a disability asset. Five demonstration projects tested novel linkage approaches, data flows and access models to inform technical aspects of the design. We adopted an iterative try-test-learn approach and forged consensus through extensive dialogue across all stakeholders to collaboratively design a system for enduring data linkage. The final design was reviewed by external experts to ensure best practice and build trust. Results A national disability-focused data asset requires linking data across all human services portfolios, as will be required for most national assets seeking to improve outcomes for particular communities or in particular domains. The proposed approach is three building blocks: national data linkage and integration infrastructure, national data system governance, and streamlined data sharing agreements for multiple uses. Current project-by-project approaches to linkage are proposed to be replaced with new enduring national keys to connect disparate state-based infrastructure and enable streamlined, reusable data curation processes. Collaboration between jurisdictions will be embedded through joint governance of the new infrastructure. Finally, a new approach to data sharing agreements between Australian governments will be developed to create safe parameters for streamlined access within existing legislation and authorisation pathways. Conclusion By designing to meet information needs across human services, this approach can be leveraged for multiple linked, longitudinal enduring data asset, beyond disability. It multiplies the benefits of upfront efforts to share and improve data, reflects a shared understanding among stakeholders, and moves Australia towards scaled use of people-centred data.

The Population Health Research Network (PHRN) is an Australian national data linkage infrastructure that links a wide range of health and human services data in privacy-preserving ways. The data linkage infrastructure enables researchers to apply for access to routinely collected, linked, administrative data from the six states and two territories which make up the Commonwealth of Australia, as well as data collected by the Australian Government. The PHRN is a distributed network where data is collected and managed at the respective jurisdictional and/or cross-jurisdictional levels. As a result, access to linked data from multiple jurisdictions requires complex approval processes. This paper describes Australia's approach to enabling access to linked data from multiple jurisdictions. It covers the identification of, and agreement to, a minimum set of data items to be included in a unified national application form, the development and implementation of a national online application system and the harmonisation of business processes for cross-jurisdictional research projects. Utilisation of the online application system and the ongoing challenges of data linkage across jurisdictions are discussed. Changes to the data custodian and ethics committee approval criteria were out of scope for this project.

Administrative health data is recognized for its value for conducting population-based research that has contributed to numerous improvements in health. In Canada, each province and territory is responsible for administering its own publicly funded health care program, which has resulted in multiple sets of administrative health data. Challenges to using these data within each of these jurisdictions have been identified, which are further amplified when the research involves more than one jurisdiction. The benefits to conducting multi-jurisdictional studies has been recognized by the Canadian Institutes of Health Research (CIHR), which issued a call in 2017 for proposals that address the challenges. The grant led to the creation of Health Data Research Network Canada (HDRN), with a vision is to establish a distributed network that facilitates and accelerates multi-jurisdictional research in Canada. HDRN received funding for seven years that will be used to support the objectives and activities of an initiative called the Strategy for Patient-Oriented Research Canadian Data Platform (SPOR-CDP). In this paper, we describe the challenges that researchers face while using, or considering using, administrative health data to conduct multi-jurisdictional research and the various ways that the SPOR-CDP will attempt to address them. Our objective is to assist other groups facing similar challenges associated with undertaking multi-jurisdictional research.