AI4DA in Healthcare


The Canadian healthcare system stands as a critical and intricate sector, greatly influencing the daily lives and well-being of millions of Canadians. Yet it is currently facing many challenges, including extended wait times, staffing deficiencies, resource allocation issues, fragmented care coordination, health disparities, and soaring costs.

At the AI Centre for Decision Analytics, we spearhead the utilization of AI, machine learning, and optimization techniques to address the complex real-world issues impacting healthcare operations today.

At a Glance

Health Expenditure

Canada spends over $300 billion annually on healthcare, accounting for about 12% of its GDP. Health expenditure has more than doubled since 2005 and is projected to continue rising.

Health Workforce

Canada’s healthcare workforce includes doctors, nurses, and other health professionals, but shortages remain a major challenge, especially in rural areas.

Hospital Capacity

Canada has about 1,300 hospitals with 2.5 beds per 1,000 people. Many hospitals operate at or near full capacity as demand for services continues to rise.

Wait Times

Canada continues to face long wait times for healthcare services, especially elective procedures and specialist consultations.

How AI Can Help

Resource Allocation

Data analysis can help optimize resource allocation, such as determining the most efficient way to schedule medical staff, allocate hospital beds, and manage operating rooms. Machine learning can assist in predicting patient demand and staffing needs based on historical data.

Patient Scheduling

Data analysis can optimize patient scheduling, reducing wait times and ensuring that healthcare providers are efficiently utilized. It can also analyze patient data to predict appointment no-shows and enable better scheduling strategies.

Supply Chain Management

Healthcare facilities rely on a steady supply of medications, medical equipment, and other resources. Data analysis can optimize inventory management, helping prevent shortages or excessive waste.

Optimizing Clinical Trials

Data analysis and machine learning can optimize the design of clinical trials, helping researchers make better use of resources and conduct trials more efficiently.

Bed Management

Data analysis can help hospitals manage bed allocation more efficiently. Machine learning can predict patient admissions and discharges, allowing for proactive bed assignment and reduced wait times.

Current Projects:

Image description

CONTEXTUAL STOCHASTIC OPTIMIZATION OF SURGERY SCHEDULING

This project builds a practical scheduling framework for elective surgeries that explicitly models uncertainty in surgery durations. Using historical operating room data, we learn context-specific duration distributions from features such as procedure type, surgeon, and patient factors. These probabilistic predictions feed a stochastic optimization model that schedules daily OR slates under real operational constraints—block times, staffing, equipment, and turnover—while balancing utilization against risks like overtime, delays, and cancellations.

To ensure deployable, we emphasize methods that are accurate, scalable, and easy to integrate into existing decision workflows (predict-then-optimize and integrated learning-and-optimization variants, with decomposition to handle large instances). The result is a scheduling tool that provides robust, data-driven plans and clear trade-offs for managers and clinicians. Our goal is to deliver improvements in OR utilization without compromising reliability, and to make the approach acceptable and actionable in real hospital settings.