New Brunswick Hospital AI Predicts Patient Discharge Times to Ease Overcrowding
Researchers at Université de Moncton have developed a predictive AI program that forecasts how long patients will need hospital care, based on historical medical data. The Vitalité Health Network, which operates New Brunswick's francophone hospitals, is testing the system to manage bed availability and reduce emergency department delays.
The province's hospitals are running at 95% occupancy, well above the safe target of 85%. More than a third of patients currently in acute care beds could receive treatment in community settings, but those services don't exist in many areas, creating bottlenecks that ripple through the system.
How the System Works
The AI examines decades of anonymized patient records-admission dates, symptoms, medical history, and discharge timing-to identify patterns. When a new patient arrives, the algorithm predicts their likely discharge date and adjusts the forecast as doctors order tests or new symptoms emerge.
Moulay Akhloufi, the computer science professor leading the project, said early results show the AI predicts discharge times as accurately as a nurse with 30 years of experience. The system is more consistent because it considers all available data without the time pressure that affects human decision-making.
"This tool is not to replace a person," Akhloufi said. "It's a tool to help the person."
What Managers Need to Know
For hospital administrators and operations staff, the value is straightforward: better visibility into bed availability allows earlier discharge planning and more efficient resource allocation. Jenny Toussaint, vice-president of clinical logistics for Vitalité, said the system helps staff plan for discharge the moment a patient occupies a bed, rather than waiting until discharge becomes urgent.
The health network shared only anonymized data with researchers. Patients are identified by numbers in the AI system, and hospitals can only connect those numbers to actual patients using their own internal records.
Timeline for Implementation
The program is still in development. The research team plans to tailor the algorithm for Dr. Georges-L.-Dumont University Hospital Centre in Moncton, where a pilot program will test real-world performance. If successful, Vitalité intends to expand it across its network.
For managers overseeing hospital operations or bed management, understanding AI for Management tools like this one is increasingly relevant to addressing capacity constraints. Learn more about AI for Healthcare applications in operational settings.
Your membership also unlocks:
