AI scribes are helping thousands of clinicians reclaim time spent on documentation. The question is can AI do more than listen and document by helping close the persistent gap between the latest medical evidence and daily clinical practice?
Historically, it takes about 17 years for evidence-based practices to be consistently implemented, with up to 40% of patients unlikely to receive evidence-based care, 1 a discrepancy with significant implications for patient outcomes and system efficiency.
While AI scribes have solved a “process” problem by reducing administrative burnout, AI-powered clinical decision support (AI-CDS) tools aim to solve a “knowledge” problem—providing real-time, context-specific guidance that reduces cognitive overload and helps clinicians navigate increasingly complex care.
However, evidence of the efficacy and adoption of AI-CDS tools remains limited. While 62% of Ontarians and 47% of clinicians report comfort with AI in health care, only 7% of clinicians currently use AI tools in patient care. 2 This underscores the need to substantiate the value of AI and address adoption barriers, particularly regarding legal responsibility and trust, to fully unlock its benefits for clinicians and patients.
The 2026 OMD Clinician Survey: AI Use in Ontario provides a more specific view of AI-CDS adoption in Ontario, with 18% of respondents (Ontario clinicians) reporting use of AI-CDS tools. Also, the survey clarifies reasons for limited uptake among clinicians who have not yet used AI-CDS tools: unavailability in EMRs (37%), lack of clarity on added clinical value (28%), lack of time (27%) and privacy concerns (27%).
A Two-Phase Evaluation Approach
Ontario Health’s Evidence2Practice (E2P) is a provincial initiative that aims to narrow the gap between clinical innovation and everyday practice by embedding guidance into front-line clinical systems in primary care and hospital settings.
In mid-2025, E2P launched a project to explore and evaluate AI-CDS tools, with partners, OntarioMD (OMD), the Centre for Effective Practice (CEP), and the Women’s College Hospital Institute for Health System Solutions and Virtual Care (WIHV). The first phase of the project, Discovery, focused on establishing foundational components to support testing activities, including identifying available AI-CDS tools on the market, conducting a systematic review of their effectiveness, examining legal and regulatory considerations for using them, analyzing the value proposition for clinicians, and developing a standardized clinical knowledge base to align relevant solutions.
Initial findings from the market analysis identified 51 relevant AI-CDS solutions for primary care. Detailed vendor information has been compiled to assess the value of these solutions for clinicians.
Physicians (n=10) engaged in the review of the value proposition felt that AI-CDS tools may be best applied primarily to administrative tasks and preventive care and diagnostic or treatment support as a secondary option. Adoption was viewed as contingent on the tools’ ability to integrate with workflows (particularly related to EMRs), explainable outputs with transparent evidence sources, and a “seek” model for diagnostic or treatment support with “nudge” reminders as preventive measures. Key risks included the deterioration of skilled labour due to automation and medicolegal liability, reinforcing the need for clear accountability and policy frameworks.
The second phase of this E2P project, known as Proof of Concept, will evaluate a representative sample of market-available AI-CDS tools, initially in a lab-based simulated environment followed by real-world testing in medical clinics across Ontario. The plan is to issue an invitation for Ontario clinicians to participate in this evaluation phase later this year.
Legal and Regulatory Analysis
The deployment of AI-CDS tools in primary care has important legal and regulatory considerations that must be addressed to mitigate risk. One key project workstream, led by OMD, is an analysis of the legislative and regulatory considerations related to AI-CDS tools. Over the past few months, our team has engaged with health system partners to examine the current legislative framework and requirements imposed by the Personal Health Information Protection Act (PHIPA) and other applicable laws together with guidance from professional regulatory bodies and associations such as the College of Physicians and Surgeons of Ontario (CPSO) and the Canadian Medical Protective Association (CMPA).
The Discovery phase identifies some issues that require further examination. There should be an aligned and strategic approach to addressing AI safety, consent, and liability issues and validation methods.
The upcoming Proof of Concept phase will be instrumental in refining these insights. By evaluating AI-CDS solutions in lab-based and clinical settings, OMD will validate risk mitigation strategies and develop recommendations. This evidence-based approach will help ensure that such strategies and recommendations are built into the potential future spread and scale of AI-CDS in Ontario.
To learn more about the AI-CDS project and other AI initiatives, sign up for OMD’s Digital Health eTips newsletter or visit our Practice Hub or website for details.
