Neil Campbell, JD is a lawyer based in Calgary, Canada with deep experience in building collaborations for research and technology development.
From my perch in the University research world, I have been privileged to witness the dramatic growth in clinical research studies testing artificial intelligence (“AI”) tools for healthcare applications. This technology holds great potential for assisting with patient diagnosis and treatment and will also be incredibly useful for generating insights across entire populations.
It’s also been exciting to see that, in some cases, preliminary test results were strong enough to lead the researchers to pursue patent protection and start the process of commercial development.
But small scale testing data cannot support the future development of any emerging AI tool destined for commercialization. Massive amounts of rich and varied data are needed to train the algorithms and create dependable tools suited for use across large populations.
In Canada, it’s currently very difficult to access population-wide data resources. The principal reason is that our data is siloed in different jurisdictions. The administration of health care services falls under the constitutional authority of ten provinces and three territories. Each jurisdiction in Canada maintains its own health records, similar in purpose but formatted differently, and for that reason records from one province cannot be easily matched with records from other provinces. (While that is a significant barrier in Canada, it’s an even bigger problem in the US where healthcare services are delivered through a vast patchwork of different providers and payors across 50 jurisdictions!)
The other big barrier is privacy legislation. All service providers in Canada’s healthcare system are governed by provincial privacy legislation designed to control the disclosure of personal health information. While the objective of ensuring patient privacy is necessary and worthy, implementing that objective led to making healthcare authorities the gatekeepers of that data, tightly controlling the use of the data – even to the extent that individuals have difficulty accessing their own data. Options for using that information outside of conventional treatment are very limited. The rigidity resulting from this structure has made it extremely difficult for population-wide data to be gathered for research and analysis by those outside each individual provincial health care system.
Concern about these barriers to the collection and disclosure of nation-wide health data have been simmering in the research and healthcare communities for years, and there have been previous attempts to tackle the problems. But with the onset of the COVID 19 pandemic, Canadians have now witnessed, with alarm, the severe negative consequences of our inability to assemble nation-wide data. The absence of this national data has led to uncertainty and delay in implementing a national response to the pandemic and will hamper our efforts to prepare for the next public health emergency.
In response, the federal government launched an initiative in 2020 to create a pan-Canadian Health Data Strategy. The first step was to appoint an Expert Panel charged with developing a plan to move forward. The Expert Panel is now mid-way through its process. In November 2021, the 2nd (of 3) reports of the Panel was released with a call for wholescale reform of the management of healthcare data in Canada. The Panel outlines specific steps to be taken jointly by the federal government, provinces and territories to remove specific barriers preventing the collection and disclosure of population-wide health data.
See: Expert Advisory Group Report 2: Building Canada's Health Data Foundation - Canada.ca
There are many reasons to be excited about the recommendations being made by the Expert Panel, but I am especially enthusiastic to see the specific endorsement of the use of population scale data for the development of new technologies, and specifically for the development of AI. If this objective is ultimately embraced by Canadian governments, it will be a huge gain for scientists and their commercial partners who are seeking to develop AI tools in the healthcare field.
If a national consensus emerges in support of the recommendations of the Expert Panel, it will still take many years to implement all the necessary structural changes. Is there any hope for major improvements in the near term? It’s too soon to say, but given the urgent pandemic-related health issues faced by Canadians, I’m guessing that those leading the initiative will feel some urgency to find interim measures to speed up the consolidation of population-wide data.
In any event, I’ll be watching for any indication that roadblocks are being removed to the aggregation of national health data for the development of commercial AI tools.
Canada is a world leader in AI research, and healthcare is one field where AI can have a huge positive impact. I’m very hopeful that big changes will be made in the near future to overhaul Canada’s health data collection systems, for the safety and well-being of Canadians, but also enabling the use of large datasets for research and product development.
.jpg)
