How a project led by Unity Health and U of T researchers could reshape Canadian health care

As medical residents at the University of Toronto, Amol Verma and Fahad Razak were struck by how many hospital decisions about quality-improvement were made based on opinion or whoever was the most persuasive voice in a room.

“They didn’t necessarily have the data to back that up,” Verma says. 

So, the pair set to work with their colleagues across the university to build a data network housed at Unity Health Toronto that linked together dozens of hospitals in Ontario, saving the province tens of millions in health-care costs, freeing up hospital beds and preventing unnecessary deaths – all while safeguarding patients’ privacy.

What began as an effort to improve decision-making in Ontario hospitals has since evolved into a national health data network called Vital that recently secured $100 million in federal support through Canada’s national AI strategy.

The milestone marks a major step for Verma and Razak – both scientists at Unity Health – as their made-in-Ontario model is poised to expand beyond its current footprint of 160 hospitals across Ontario, Quebec and Alberta.

Vital will allow researchers to harness anonymized data to improve patient care and conduct large-scale research projects that will ultimately save lives. Ultimately, there are plans to expand the network, based at Unity Health’s St. Michael’s Hospital, to encompass all 13 of Canada’s provinces and territories – with seven already expressing interest and readiness to join the network.

“U of T is proud of the visionary leadership of Amol Verma and Fahad Razak, who have drawn on the highly collaborative ecosystem of researchers, hospital partners and leading AI institutes in the Toronto region to build a tool that is having a transformative impact on health care in Ontario,” said Leah Cowen, U of T’s vice-president, research and innovation, and strategic initiatives. 

“With support from Canada’s national AI strategy, Vital is now poised to scale that impact across the country, advancing a bold vision for data-driven health care and reinforcing U of T’s commitment to world-leading innovation in AI.”

As Vital co-leads, Verma and Razak are overseeing the scale up of the model they co-founded in Ontario more than a decade ago. Anonymized data from that platform, called GEMINI, has so far been used to support more than 150 projects involving research projects. That includes: Canada’s largest AI-focused clinical trial, which is focused on predicting and preventing delirium in patients; developing and implementing a tool to predict limb amputations in diabetes and flag patients for early prevention; and a study focused on improving the prediction of sepsis – a leading cause of hospital deaths. 

Reducing wait times, saving money

“Through partnership with the Ontario government and significant investment over the last 10 years, physicians and hospital executives now get detailed information about general medicine hospital care, statistically benchmarked, which has led to meaningful improvements in health-care delivery and quality in the province,” said Razak, an associate professor of medicine in U of T’s Temerty Faculty of Medicine and the Canada Research Chair in data-informed health-care improvement. 

“Through the use of our analysis and reporting, we've reduced 50,000 patient-days in hospital across our network over the last 18 months.”

That translates into tens of millions in savings for the province, Razak added, while helping patients move more quickly through the hospital system and into appropriate care.

In Ontario, GEMINI’s work has been made possible by deep collaborations across the sprawling health sciences ecosystem in Toronto – namely the Toronto Academic Health Sciences Network(TAHSN), which includes U of T and 14 partner hospitals. It also draws on the expertise of researchers at U of T’s Data Sciences Institute, an institutional strategic initiative, Temerty Centre for AI Research and Education in Medicine (T‑CAIREM) and Vector Institute.

The value of this approach became especially clear during the COVID-19 pandemic, when GEMINI supported more than 20 studies that helped hospitals understand disease severity, anticipate drug shortages and identify disparities in care.

“If we can bring this together at a national scale in Canada, we can create a scientific data repository that is among the best in the world,” said Verma, an associate professor in Temerty Medicine’s department of laboratory medicine and pathobiology and the Temerty Professor in AI Research and Education in Medicine.

Faster, more equitable health care

The pandemic also exposed a critical limitation: hospital data can often take several months to get shared, limiting its usefulness.

Reducing that lag time is a central goal of Vital. The aim is to move toward data access within 24 hours, which would allow more timely responses to emerging health threats. It would also bring faster, large-scale clinical trials to Canada – providing patients with access to novel treatments sooner while attracting research investment.

Ensuring equitable access to these innovative, cutting-edge treatments is another key priority. 

“Right now, if you have the benefit of living in downtown Toronto or Ottawa, you get access to the latest clinical trials or other innovation … in a way that someone in Sarnia or Parry Sound would not,” said Razak, who holds a Canada Research Chair in Healthcare Data and Analytics. “So, we have wildly unequal access to new innovations in Canada and we leave our rural and northern populations behind.”

Altaf Stationwala, president and CEO of Unity Health, said inclusive data leads to more equitable health-care solutions. 

“We’re proud to be at the forefront of this initiative which will strengthen Canada’s ability to lead the world in health care innovation,” Stationwala said.

Vital is also being developed with a strong focus on Indigenous data sovereignty and partnership, including collaborations with Indigenous-led organizations, including the Well-Living House, to ensure research is governed according to community priorities, helping build trust in how health data is used. The Well Living House is led by Métis physician Janet Smylie, a clinician-scientist at Unity Health and professor at U of T’s Dalla Lana School of Public Health.

Realizing Vital’s vision at a national scale requires new ways of working together across Canada’s provincially run health systems.

“Historically, health care has been an area where it’s been difficult for provinces to collaborate,” Verma said.

Safeguarding privacy

To address this, Vital uses a federated data model that allows data to remain controlled by the provinces while still being analyzed collectively – an approach that enables collaboration without compromising privacy. The expertise at the Data Sciences Institute and T-CAIREM are crucial in this regard. “U of T’s strength in biostatistics and machine learning makes it a world-leading place to develop the tools that enable this kind of work,” Verma said.

He added that the initiative will also play a key role in developing future leaders in the rapidly advancing field of AI-powered health care.

“In helping attract some of the best talent in the world to do research and generate innovations in this area, Vital will make a major contribution to Canada and our health-care system.”

GEMINI has also partnered with the Vector Institute to support the development and testing of AI tools using real-world clinical data – an approach that has received the backing of Geoffrey Hinton, University Professor emeritus of computer science and winner of a 2024 Nobel Prize in Physics who is regarded as the “godfather of AI.”

“Canada, due to its diverse population, open society and inclusive healthcare system, coupled with AI talent, can lead health AI innovation and safety research. But this has been stymied by a lack of access to deep clinical data with good governance,” said Hinton. 

“Vital will overcome these barriers and allow Canadians to fully benefit from this transformative technology.”