A pilot project testing an artificial intelligence-enabled clinical decision support (AI-CDS) tool has demonstrated the potential of emerging technology to help primary care clinicians improve the prevention, screening, monitoring, and treatment of Type 2 diabetes.
Type 2 diabetes remains one of the most common and complex chronic diseases managed in primary care. Effective care requires ongoing monitoring, timely screening, medication reviews, preventive interventions, and adherence to evidence-based clinical guidelines. At the same time, family doctors and nurse practitioners face increasing demands during patient visits, often navigating large amounts of information within electronic medical records (EMRs) while making time-sensitive clinical decisions.
Recognizing these challenges, the pilot project explored whether an AI-CDS tool could help reduce clinicians' cognitive burden and support more consistent, guideline-informed care for patients with or at risk of Type 2 diabetes.
Testing AI-CDS in Community Primary Care
The project provided an opportunity to evaluate the use of the
Healwell AI Clinical Decision Support System (CDSS) in three family practices that are partners of the
Mid-West Toronto Ontario Health Team (MWT-OHT): the
Mid-West Toronto Family Practice Network and the
Toronto Western Family Health Team - University Health Network. Fourteen primary care clinicians participated in the pilot, which involved nearly 500 patients over a period of almost three months.
The AI-CDS tool was designed to analyze patient information from the EMR and match it with targeted clinical knowledge based on Ontario's Quality Standards. Recommendations were presented to clinicians before and during patient visits, helping them identify care gaps and make informed decisions as part of their existing workflow.
The tool supported clinicians by highlighting actions such as when to order laboratory tests, review medications, refer patients to specialists or schedule follow-up appointments. Clinicians always retained the final authority over patient care decisions and determined whether to accept or act on the tool's recommendations.
Collaborative Effort Across Multiple Partners
The pilot was developed by the MWT-OHT and brought together several organizations with complementary expertise.
The AI-CDS technology used during the project was the Healwell AI CDSS. The
Centre for Effective Practice (CEP) supplied the clinical guidelines that informed the tool's recommendations. OntarioMD contributed expert advice on quality improvement measures and conducted the end-of-project evaluation.
Boehringer Ingelheim coordinated project partners, provided implementation expertise, and funded the initiative.
Together, the partners aimed to assess whether AI-enabled decision support could be effectively implemented in community primary care settings and generate meaningful improvements in diabetes care.
Key Findings and Early Benefits
The pilot demonstrated that AI-CDS technology has practical applications in real-world primary care environments. While participants acknowledged that the technology still requires refinement, the project showed that AI-CDS can move beyond theory and be successfully implemented as a pilot in community practices.
Among the most notable findings, the AI-CDS tool helped identify patients at high risk of developing Type 2 diabetes and highlighted opportunities to improve care for patients already living with the condition. Clinicians reported that the technology helped them find patients who needed additional attention and follow-up, supporting more proactive disease management.
The project also enhanced clinicians' experience by integrating AI-enabled decision support directly with EMR data. By surfacing relevant clinical information and recommendations, the tool aimed to reduce cognitive load, improve consistency in care delivery, increase efficiency, and support higher-quality patient care.
Participating clinicians reported value in using the tool to support routine care and follow-up activities for patients with Type 2 diabetes. Many also recognized the potential for similar AI-CDS tools to be applied to other chronic diseases in the future.
Lessons Learned for Future Adoption
Despite the encouraging results, the pilot also revealed important challenges that must be addressed before broader adoption.
Some clinicians reported that the tool slowed their workflow and did not reduce administrative burden. In some cases, it added work to already busy practices. Concerns were also raised about technical reliability and system performance, highlighting the importance of ensuring stable and dependable technology in clinical settings.
Participants also identified opportunities to improve the accuracy and usability of the tool. Suggested enhancements included better integration into existing clinical workflows, reducing administrative demands, resolving technical issues, and refining the algorithms to reduce false-positive diabetes alerts and increase clinician confidence in the recommendations.
Looking Ahead
The lessons learned from this pilot provide valuable insights for the future development and implementation of AI-CDS technologies in Ontario primary care. More details will be available in an upcoming formal peer-reviewed journal article.
The project demonstrated that AI-enabled clinical decision support can help clinicians identify care opportunities, support evidence-based decision-making and improve follow-up for patients with Type 2 diabetes. At the same time, the findings highlight the importance of designing tools that fit seamlessly into practice workflows and deliver a reliable user experience.
As AI technology continues to evolve, the experience gained through this pilot will help inform future improvements and support decisions about wider adoption. With ongoing refinement, AI-CDS tools have the potential to become an important part of primary care, helping clinicians deliver more proactive, consistent, and high-quality care for patients living with chronic diseases such as Type 2 diabetes.