A large-scale clinical trial launching across 13 Ontario hospitals will use artificial intelligence to identify patients at high risk of delirium, a condition that affects roughly one in four hospitalized adults. The trial, called AI Models (AIM) to Prevent Delirium, will run through March 31, 2027, and involve approximately 15,000 patients.
Delirium is a state of new-onset confusion that can cause agitation and disorientation. It affects about 500,000 Canadians annually and is linked to increased risk of dementia and mortality. The condition also adds an estimated $11,000 in extra costs per hospital admission due to extended stays.
The project is funded by Ontario's Ministry of Health in partnership with Ontario Health. It is led by the GEMINI research network at Unity Health Toronto, in collaboration with the University of Toronto, the Centre for Quality Improvement and Patient Safety, the Regional Geriatric Program of Toronto, and the Toronto Academic Health Sciences Network.
How the AI model works
Up to 40 per cent of delirium cases are preventable, but prevention requires frequent monitoring and interventions addressing patients' medical conditions, nutrition, sleep, and mental state. That level of care is difficult to sustain under real-world staffing conditions.
The AI model uses just 10 routinely available factors - such as age, medical conditions, and standard lab tests - to predict delirium risk with 70 per cent accuracy. The engineering team, led by Professor Eldan Cohen at the University of Toronto, kept the number of inputs small so the tool can work at any hospital, not just those with advanced IT systems.
"Advanced engineering methods can help us create usable and accessible health AI solutions. The mathematical equations within the model are still quite complex, but the number of inputs is small," said Cohen.
The minimal data requirements also protect patient privacy, as the model does not need any identifying information.
Co-designing the intervention with care teams
Over the past year, teams at all participating hospitals have been co-designing the intervention so delirium prevention efforts fit into regular workflows. The trial will evaluate the program across different organizations to create tools ready to scale widely.
"It has been inspiring to see nurses, doctors, patients and caregivers, hospital administrators, and quality improvement specialists come together to design and implement this intervention," said Dr. Brian Wong, Director of the Centre for Quality Improvement and Patient Safety.
Dr. Amol Verma, the project's principal investigator, said the trial represents a partnership between AI scientists, front-line teams, patients, and policymakers. "Artificial Intelligence offers many promising opportunities in healthcare, but we need to ensure it is used safely and effectively," he said.
Patient partner Nicole Lafreniere-Davis described the human cost of delirium: "In addition to the stress hospitalization was already causing, my husband's delirium episode triggered fear, confusion, sadness, not just for him, but for our entire family."
Participating hospitals
The 13 sites include Humber River Health, London Health Sciences Centre (University and Victoria hospitals), North York General Hospital, Scarborough Health Network (General and Birchmount), Unity Health Toronto (St. Joseph's and St. Michael's), Sunnybrook Health Sciences Centre, University Health Network's Toronto Western, Trillium Health Partners (Credit Valley and Mississauga), and Niagara Health's Marotta Family Hospital.
GEMINI, the research network behind the project, is Canada's largest hospital data-sharing network for research. It collects standardized, de-identified clinical data from more than 40 Ontario hospitals, covering 60 per cent of inpatient adult medicine care and 70 per cent of paediatric care. The network has supported more than 150 research projects.
Why this matters for healthcare professionals
For clinicians and hospital administrators, this trial tests whether AI can help prioritize care under real-world constraints. Delirium prevention is known to work, but it has been difficult to implement consistently. If the model succeeds, it could give care teams a practical way to direct prevention efforts to the patients who need them most - without requiring new data infrastructure or additional staffing.
Your membership also unlocks:
