Funding supports AI-powered clinical decision-making for diabetes care

People living with diabetes often take several medications and need regular monitoring. When choosing or adjusting a treatment, health professionals must weigh clinical guidelines against laboratory results and each patient’s circumstances.

CONNECT aims to make that task easier by bringing together data on prescribed medications and diabetes care in a structured, searchable resource.

Three partners will contribute. Omnimed will provide data from electronic health records. Optithera will contribute genetic risk information, while Wikimedica will supply drug knowledge and clinical care pathways. Common standards will make the information easier to use across systems and validate.

How will the tool work? Given a clinical scenario, it will first retrieve relevant passages from guidelines and available evidence. A large language model will then help draft a structured suggestion, such as a medication adjustment or a test to consider.

Each suggestion will be linked to its supporting sources, allowing clinicians to review the evidence behind it.

“Canada needs more than great ideas, we need a trusted ecosystem where advanced AI algorithms can be tested, compared, and proven on real-world data under strong governance,” said Nikiema.

Keeping patient data in place

Partners will not need to exchange individual patient records to work together. Models will be trained and used where the data already reside. Only encrypted model updates and aggregated statistics will move between organizations.

This method, known as federated learning, is central to the national platform. It enables analysis of data held across multiple institutions while respecting privacy and the rules governing how the information is used.

The CONNECT team will also validate and compare several large language models. It then plans to integrate a decision-support prototype into Omnimed’s electronic health record system, giving health professionals access to the tool within their existing workflow.

The project is therefore expected to deliver better-organized data, an evaluation of the models and a prototype for clinical use. The goal is to support safer, more consistent decisions while accounting for the needs of each person living with diabetes.