The Roundtable session of the CHIEF Executive Forum Spring Symposium focused on turning AI principles into actionable strategies for healthcare. How can AI implementation be accelerated and what is the role of governments in scaling AI initiatives? Following are concrete actions needed to advance AI adoption for healthcare as identified by the roundtable participants to encourage development of a cohesive national vision for AI in healthcare—aligning policies, strategies, and goals across regions and sectors to ensure a unified approach to AI adoption and integration.
Define challenges and priorities.
Co-design AI solutions with a clinical team using a bottom-up approach. Look at the pain points and determine if AI is appropriate. Establish requirements and design from these. Move to proof of concept, and then can scale and spread. Be agile when responding to end-user feedback. Develop multiple iterations of AI solutions.
Accelerate the implementation of low-risk (e.g., no PHI) high-impact scenarios. Use AI (such as a Digital Twin or Digital Front Door) to identify areas requiring attention. Set priorities for AI development and adoption using Canada’s four shared health priorities.
Create a platform for sharing models and use cases.
Establish a centralized repository or organization to coordinate efforts, share resources, and avoid duplication of work in AI development and implementation. Share trained AI models across provinces. Create a centralized data hub which allows for data to be published and become accessible to enable more AI models. Facilitate the sharing of learnings and experiences from various AI implementations to foster continuous improvement. Encourage outsourcing and resource sharing to address limited capacity for AI expertise.
Educate and communicate what can and can’t be done with AI (education and implementation should be completed in parallel).
Define terms surrounding AI to mitigate differences in interpretations. Identify clinical champions and bring clinicians into early stages of design. Implement standardized training programs to educate healthcare professionals, administrators, and patients about AI technologies, ensuring widespread understanding and proper utilization. Build patient engagement. Highlight and promote successful AI use cases to demonstrate the benefits of AI.
Ensure equity and access.
Collect equity data to provide equitable care for all patients. Standardize methods for data collection. Utilize maturity models. Democratization with Large Language Models (LLMs), and Intellectual Property (IP) clauses. Ensure transparency in AI use to maintain trust and ethical standards.
Build data strategies.
Maintain focus on improving access to and sharing of quality data, and use AI to help close gaps and get the data captured and sorted (i.e., pull from EMR notes field into IPS). Declare the data: what dataset is being used to train models to increase transparency and highlight any potential biases in the model? Ensure high quality and standardization of data (create a common dictionary or lexicon). Use data that includes human workflow considerations.
Build data governance with AI standards and meaningful regulations.
Develop a common framework for AI implementation. Provide guidelines and best practices for ensuring the quality and reliability of AI systems. Employ evidence-based standards. Develop AI certification and accreditation models to standardize and ensure the quality and safety of AI applications. Treat algorithms as medical devices (as in some international jurisdictions). Balance the need for regulation to ensure patient safety and data security with the imperative to not stifle innovation. Categorizing AI projects by risk level to determine the appropriate level of regulatory scrutiny and support. Invest locally in redesigning healthcare systems to accommodate AI integration. Publish a list of government approved use cases. Ensure the list is flexible and not restrictive, allowing for expansion as AI technology evolves.
Develop a comprehensive framework to manage risks associated with the deployment of AI in healthcare, ensuring patient safety and ethical standards are maintained.
Conduct regular risk assessments to identify and mitigate potential issues with AI systems. Align AI projects with organizational vision. Identify and focus on vertical use cases rather than generic tools.
Establish robust partnerships with vendors to help address talent retention challenges and ensure that AI solutions are tailored to meet specific local needs. Build test environments with vendors to trial AI solutions. Be transparent with any partnership (and understand the competition). Explore collaborations with research organizations and establish relationships. Bring industry together to gather data and evidence, and then go to government with the reports.
