This study explores how patients, caregivers, and community stakeholders can be meaningfully involved throughout the life‑cycle of artificial‑intelligence (AI) tools for population‑health interventions. It asks:
Which principles should guide community engagement in AI‑driven diabetes prevention?
How can those principles be operationalized in practice?
The approach, questions, findings, and recommendations are relevant well beyond the specific scope of population health and technology.
What is this research about?
This research provides practical, principled guidance for embedding meaningful, equitable community engagement into AI initiatives aimed at improving population health outcomes like diabetes prevention.
This research is about identifying principles and practices for meaningful community engagement in the deployment of artificial intelligence (AI) tools to prevent and manage type 2 diabetes (T2D) at a population health level.
It aimed to answer guiding questions on how patients, caregivers, and communities can be engaged responsibly and effectively in the AI life cycle stages to ensure socially acceptable and equitable AI solutions.
What do you need to know?
Diabetes is a major global public health issue with increasing prevalence and structural barriers to prevention.
AI, especially machine learning, has predictive power for T2D onset and complications, but adoption in public health is slow partly due to lack of community engagement.
Current AI health research focuses on governance and technical optimization but rarely involves patient/community perspectives, limiting social acceptability and equity.
This study is unique in combining a literature scan and a participatory workshop with diverse stakeholders including people with lived experience of T2D to identify engagement principles spanning the entire AI life cycle (define, design/test, deploy, evaluate, improve).
The research advances understanding by integrating the International Association for Public Participation (IAP2) engagement spectrum with the AI life cycle stages to create a novel conceptual framework for community engagement in AI.
What did the researchers find?
From literature, 10 principles for community engagement in AI were identified: trust, power-sharing, empowerment, value alignment, equity, codesign, transparency, education, early engagement, accountability.
Workshop participants (n=30) ranked the top 6 as trust, equity, accountability, transparency, codesign, and value alignment.
A conceptual framework was developed showing how engagement can occur at any AI life cycle stage and at varying levels of participation (inform, consult, involve, collaborate, empower).
Operational practices for each top principle were co-developed, such as engaging trusted community leaders for trust; ensuring diversity and cultural responsiveness for equity; establishing community co-leadership and ongoing communication for accountability; honest and accessible information sharing for transparency; continuous iterative feedback for codesign; and aligning AI goals with community values through dialog.
Participants stressed inclusivity, sustainability, departure from one-off engagement to ongoing centralized models, and culturally appropriate communication methods.
Quotes include emphasis on trust through “repeated engagement with clear, honest, and consistent 2-way communication” and codesign as “continuous and iterative feedback” with “no points of no return”.
What are some particularly interesting themes and outlier findings?
Trust emerged as the most critical principle, given AI’s "Blackbox" nature which heightens skepticism.
Inclusivity and equity were not just principles but also outcomes of meaningful engagement.
Education, while not independently prioritized, was foundational as a digitally literate citizenry was deemed necessary.
A community ambassador model was highlighted as a notable asset for reaching marginalized populations.
Workshop participants called for culturally tailored approaches in a diverse region, emphasizing varied accessibility needs and structural barriers.
Importantly, the study found that while community engagement ideally happens in early AI stages, many participants were not involved in problem definition or model development, highlighting an implementation gap.
How can you use this research?
AI developers should incorporate the six prioritized principles into the entire AI development and deployment process to foster trust and equity.
Public health officials and policymakers can use the conceptual framework to design inclusive engagement strategies ensuring meaningful participation of diverse community members.
Clinicians and community organizations can leverage these findings to advocate for patient and public involvement in AI-based health interventions.
Researchers can build on the study's framework and findings to refine engagement methodologies and develop community advisory groups.
Practitioners and academics are encouraged to operationalize the principles into guidelines for AI deployment in population health contexts.
The study suggests ongoing monitoring and adaptation, addressing unintended consequences and maintaining transparency for sustained social acceptability.
What did the researchers do?
They conducted a literature scan to identify existing AI and digital health community engagement strategies.
Organized a participatory workshop with 30 stakeholders including patients, caregivers, community representatives, clinicians, AI researchers, and funders in Peel Region, Ontario, a high diabetes prevalence area.
Used a formative qualitative study design with nominal group technique for structured participation and ranking exercises.
Data was collected via notes, sticky notes, and polls, then analyzed with thematic coding integrating deductive and inductive strategies, mapping to literature-identified principles.
Participants received honoraria; ethical approval was obtained from institutional review boards.
The workshop informed a conceptual framework integrating the AI life cycle stages with levels of public participation (IAP2 spectrum).
This study explores how patients, caregivers, and community stakeholders can be meaningfully involved throughout the life‑cycle of artificial‑intelligence (AI) tools for population‑health interventions.
