Establishing organizational AI governance in healthcare: a case study in Canada

May combine two claims.

Participants responded positively to the process outlined in the PPTO framework, finding it practical and aligned with the organizational needs.

“They responded positively to the process outlined in the PPTO framework20, finding it practical and aligned with the organizational needs.”

• cites [20] Adam, H. et al. Write it like you see it: detectable differences in clinical notes by race lead to differential model recommendations. In Proc. 2022 AAAI/ACM Conference on AI, Ethics, and Society 7–21 (ACM, 2022). sourceCould not verify

  The Methods (section 5.1) describe recruiting physicians for an experiment, and the Results (section 5.2) report their accuracy in predicting patient race. Searches for “PPTO,” participant feedback, and organizational needs found no passage about a PPTO framework or participants finding its process practical. (LLM, conf=0.76)

  Experimental — accuracy not yet measured

May combine two claims.

Implementation challenges for AI in clinical use include securing significant investment in funding, building infrastructure, employing skilled personnel, ensuring regulatory compliance, and integration into clinical workflows.

“It entails overcoming implementation challenges, such as securing significant investment in funding, building infrastructure, employing skilled personnel, ensuring regulatory compliance, and integration into clinical workflows14–16.”

• cites [14] Boag, W. et al. The algorithm journey map: a tangible approach to implementing AI solutions in healthcare. npj Digit. Med. 7, 87 (2024). sourcePartly supported

  The Results (Algorithm journey map; page unavailable) describe building a real-time data pipeline and server infrastructure, consulting regulatory affairs leadership, adapting clinical workflows, and periodically securing funding and resources. The Results (Interviewee-identified opportunities for improvement; page unavailable) also describe recruiting statistical expertise. These passages support the listed implementation challenges, but do not establish that the required financial investment was significant. (LLM, conf=0.83)

  Experimental — accuracy not yet measured
• cites [15] Kim, J. Y. et al. Organizational governance of emerging technologies: AI adoption in healthcare. In Proc. 2023 ACM Conference on Fairness, Accountability, and Transparency 1396–1417 (ACM, 2023). sourceSupported
• cites [16] Reddy, S., Allan, S., Coghlan, S. & Cooper, P. A governance model for the application of AI in health care. J. Am. Med. Inform. Assoc. 26, 1034–1037 (2019).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Participants found the stakeholders outlined in the PPTO framework to be relevant, with clearly defined roles and responsibilities.

“Participants found the stakeholders outlined in the PPTO framework20 to be relevant, with clearly defined roles and responsibilities.”

• cites [20] Adam, H. et al. Write it like you see it: detectable differences in clinical notes by race lead to differential model recommendations. In Proc. 2022 AAAI/ACM Conference on AI, Ethics, and Society 7–21 (ACM, 2022). sourceCould not verify

  Our copy’s Discussion and Conclusion addresses race prediction from clinical notes and its implications for care, not participants’ assessment of a stakeholder framework. Searches for “PPTO framework stakeholders roles responsibilities” and the claimed participant assessment found no passage addressing the PPTO framework or whether its roles were clearly defined. (LLM, conf=0.76)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim.

Addressing safety and ethical concerns, including algorithmic bias, data privacy and security, and transparency, is a critical part of implementing AI in healthcare.

“More critically, it involves addressing safety and ethical concerns, including algorithmic bias, data privacy and security, and transparency16,17.”

• cites [16] Reddy, S., Allan, S., Coghlan, S. & Cooper, P. A governance model for the application of AI in health care. J. Am. Med. Inform. Assoc. 26, 1034–1037 (2019).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [17] Suresh, H. & Guttag, J. A framework for understanding sources of harm throughout the machine learning life cycle. In Proc. 1st ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization 1–9 (ACM, 2021). sourcePartly supported

  The Introduction and §3.8 describe harms from ML systems and identify bias in medical applications as a concern. Section 3.5.1 discusses privacy, including a trade-off between differential privacy and performance for underrepresented data. The paper does not substantively address data security or transparency as concerns in implementing AI in healthcare, so it supports only part of the attributed list. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

The collection and use of large data sets raise concerns about privacy violations.

“The collection and the use of large data sets raise concerns about privacy violations18.”

• cites [18] Chustecki, M. Benefits and risks of AI in health care: narrative review. Interact. J. Med. Res. 13, e53616 (2024). sourceSupported

May combine two claims.

AI adoption in healthcare is expected to grow, but its actual implementation currently remains limited due to a variety of challenges.

“While AI adoption is expected to grow, its actual implementation currently remains limited due to a variety of challenges.”

• cites [7] Bajwa, J., Munir, U., Nori, A. & Williams, B. Artificial intelligence in healthcare: transforming the practice of medicine. Future Healthc. J. 8, e188–e194 (2021). sourceSupported
• cites [4] He, J. et al. The practical implementation of artificial intelligence technologies in medicine. Nat. Med. 25, 30–36 (2019). sourceSupported
• cites [13] Dreyer, K. J. & Geis, J. R. When machines think: radiology’s next frontier. Radiology 285, 713–718 (2017). sourceNot checked — no full text of the source

May combine two claims.

AI governance frameworks specifically designed for healthcare largely emphasize high-level ethical principles or focus on technical considerations related to algorithm development, implementation, and maintenance.

“However, these frameworks largely emphasize high-level ethical principles30–32 or focus on technical considerations related to algorithm development, implementation, and maintenance33,34.”

• cites [30] Abràmoff, M. D. et al. Considerations for addressing bias in artificial intelligence for health equity. npj Digit. Med. 6, 170 (2023). sourcePartly supported

  The Introduction’s ‘AI total product lifecycle’ section covers development, implementation, deployment, and monitoring, including metrics for evaluating AI systems. But it describes the paper’s framework as extending lifecycle analysis to equity and the wider healthcare context, so characterizing its focus as largely technical omits a central part of it. The supplied copy gives section labels but no page numbers. (LLM, conf=0.82)

  Experimental — accuracy not yet measured
• cites [31] Chen, I. Y. et al. Ethical machine learning in healthcare. Annu. Rev. Biomed. Data Sci. 4, 123–144 (2021). sourcePartly supported

  The Algorithm Development and Postdeployment Considerations sections discuss technical choices, deployment, and ongoing model audits, supporting this paper as an example of work addressing those stages. But the Introduction and Problem Selection section also address social justice, research funding, and whose health needs receive attention; describing the paper as focused on technical considerations alone omits that scope. The paper does not establish what healthcare AI governance frameworks largely emphasize as a group. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites [32] Chin, M. H. et al. Guiding principles to address the impact of algorithm bias on racial and ethnic disparities in health and health care. JAMA Netw. Open 6, e2345050 (2023). sourcePartly supported

  The Conceptual Framework section (page not available in our copy) covers algorithm development, deployment, monitoring, and maintenance. But the Abstract describes a broader framework centered on equity, community engagement, transparency, and accountability, so characterizing this paper as focused largely on technical considerations overstates its emphasis. (LLM, conf=0.82)

  Experimental — accuracy not yet measured
• cites [33] Coombs, L. et al. A machine learning framework supporting prospective clinical decisions applied to risk prediction in oncology. npj Digit. Med. 5, 117 (2022). sourceSupported
• cites [34] Collins, G. S. et al. TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ 385, q902 (2024). sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

Various qualitative research methods, such as surveys, interviews, and co-design workshops, can be instrumental in facilitating this collaborative process.

“Various qualitative research methods, such as surveys, interviews, and co-design workshops, can be instrumental in facilitating this collaborative process40.”

• cites [40] McAuliffe, E. et al. Collective leadership and safety cultures (Co-Lead): protocol for a mixed-methods pilot evaluation of the impact of a co-designed collective leadership intervention on team performance and safety culture in a hospital group in Ireland. BMJ Open 7, e017569 (2017). sourcePartly supported

  The Methods section (Phase 2; page not provided) proposes co-design workshops and says interview findings will inform the intervention’s co-design. The Methods section (Phase 1; page not provided) also describes a survey, but it is used to collect network data, not presented as a qualitative method facilitating co-design. The paper is a protocol, so it proposes these methods rather than demonstrating that they were instrumental. (LLM, conf=0.78)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Initial interactions among stakeholders from diverse backgrounds and areas of expertise can often be met with apprehension.

“It is especially important because initial interactions among stakeholders from diverse backgrounds and areas of expertise can often be met with apprehension41.”

• cites [41] Pallesen, K. S., Rogers, L., Anjara, S., De Brún, A. & McAuliffe, E. A qualitative evaluation of participants’ experiences of using co-design to develop a collective leadership educational intervention for health-care teams. Health Expect. 23, 358–367 (2020). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Qualitative methods are widely used to explore stakeholders' experiences, challenges, and needs in the adoption of AI.

“Qualitative methods are widely used to explore stakeholders’ experiences, challenges, and needs in the adoption of AI37–39.”

• cites [37] Leenen, J. P. L. et al. Exploring the complex nature of implementation of Artificial intelligence in clinical practice: an interview study with healthcare professionals, researchers and policy and governance experts. PLOS Digit. Health 4, e0000847 (2025). sourceSupported
• cites [38] Siira, E., Tyskbo, D. & Nygren, J. Healthcare leaders’ experiences of implementing artificial intelligence for medical history-taking and triage in Swedish primary care: an interview study. BMC Prim. Care 25, 268 (2024). sourceSupported
• cites [39] Sendak, M. et al. Building models, building capacity: a review of participatory machine learning for HIV prevention. PLOS Glob. Public Health 5, e0003862 (2025). sourcePartly supported

  Section 2.3 describes focus groups examining patients’ views of an HIV-prediction AI system and clinicians’ concerns about its implementation. The HEROIC case study in section 2.4 also reports interviews about workflow and implementation needs. These examples support the use of qualitative methods for that purpose, but do not establish that they are widely used across AI adoption generally. The indexed copy provides no page numbers for these passages. (LLM, conf=0.85)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

A hospital system co-designed a team-based leadership model in collaboration with healthcare teams and patients, using a range of qualitative methods, including in-depth interviews and online surveys.

“For example, a hospital system co-designed a team-based leadership model in collaboration with healthcare teams and patients, using a range of qualitative methods, including in-depth interviews and online surveys.”

• cites [40] McAuliffe, E. et al. Collective leadership and safety cultures (Co-Lead): protocol for a mixed-methods pilot evaluation of the impact of a co-designed collective leadership intervention on team performance and safety culture in a hospital group in Ireland. BMJ Open 7, e017569 (2017). sourcePartly supported

  Methods, Phase 2 (page unavailable) proposes co-designing a team-focused collective leadership intervention with healthcare staff and patient representatives, using workshops informed in part by interviews. But this is a protocol for planned work, not a report that the hospital system had already co-designed the model. The online survey described in Methods, Phase 1 (page unavailable) is for mapping leadership networks, not a qualitative co-design method. (LLM, conf=0.87)

  Experimental — accuracy not yet measured

May combine two claims.

Participants found technology infrastructure described in the PPTO framework relevant and agreed to use it to examine gaps in their existing system.

“Participants found technology infrastructure described in the PPTO framework20 relevant and agreed to use it to examine gaps in their existing system.”

• cites [20] Adam, H. et al. Write it like you see it: detectable differences in clinical notes by race lead to differential model recommendations. In Proc. 2022 AAAI/ACM Conference on AI, Ethics, and Society 7–21 (ACM, 2022). sourceCould not verify

  The Methods section describes a web-based experiment with physicians, while the Introduction and Discussion address racial bias in clinical notes and machine-learning systems. Searches for the PPTO framework and for participants agreeing to examine gaps in an existing system found no passage addressing either point in our copy. (LLM, conf=0.77)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

These methods have also been shown to effectively support organizational change in healthcare.

“These methods have also been shown to effectively support organizational change in healthcare.”

• cites [40] McAuliffe, E. et al. Collective leadership and safety cultures (Co-Lead): protocol for a mixed-methods pilot evaluation of the impact of a co-designed collective leadership intervention on team performance and safety culture in a hospital group in Ireland. BMJ Open 7, e017569 (2017). sourcePartly supported

  The Background and Methods describe a co-designed collective-leadership intervention intended to improve healthcare teams, while the Discussion says collaborative design increases its likelihood of success. But this is a protocol for a future evaluation: the Discussion says the intervention could be scaled up only ‘should it prove successful,’ so the paper does not show that these methods effectively support organisational change. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

May combine two claims.

Additional key stakeholders essential to AI implementation were identified, including a business committee managing the software budget, departmental leaders overseeing budget approvals, quality requirements, and technology use, clinical champions representing end users, IT teams responsible for technical infrastructure and data management, data scientists assessing AI solutions, and external AI-s

“The interviews also revealed additional key stakeholders essential to AI implementation”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Key considerations for AI governance committee member expectations should include the general time commitment expected from committee members, the availability of full-time staff to oversee the operations of AI governance, and whether committee members will be compensated for their service, along with the structure of that compensation.

“Key considerations should include the general time commitment expected from committee members, the availability of full-time staff to oversee the operations of AI governance, and whether committee members will be compensated for their service, along with the structure of that compensation.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants unanimously suggested that the membership of the AI governance committee should be reviewed and updated annually per organizational standard governance process.

“They also unanimously suggested that the membership of the AI governance committee should be reviewed and updated “annually per organizational standard governance process.””

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants noted that while technical and informatics experts would focus primarily on technical domains, they should also have a foundational understanding of clinical workflows to support socio-technical alignment.

“They also noted that while technical and informatics experts would focus primarily on technical domains, they should also have a foundational understanding of clinical workflows to support socio-technical alignment.”

May combine two claims.

The stakeholder interviews and co-design workshops led to the creation of policies and the formation of an AI governance committee within the organization.

“Together, these efforts led to the creation of policies and the formation of an AI governance committee within the organization.”

May combine two claims.

The organization possesses a robust data infrastructure and strong capabilities to support AI initiatives, with several vendor-based AI products already in place.

“The interviews revealed that the organization possesses a robust data infrastructure and strong capabilities to support AI initiatives, with several vendor-based AI products already in place.”

May combine two claims.

IT teams oversaw data de-identification and re-identification processes, ensuring secure vendor access to data while maintaining its confidentiality.

“Additionally, the teams oversaw data de-identification and re-identification processes, ensuring secure vendor access to data while maintaining its confidentiality.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants recommended adding a responsibility to the AI governance committee in general-ensuring return on investment-which was not originally included in the framework.

“Participants recommended adding a responsibility to the AI governance committee in general—“ensuring return on investment”—which was not originally included in the framework.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Participants unanimously reported that the AI governance committee should be a subcommittee reporting up to DHC, leveraging its members' expertise and facilitating smoother implementation, rather than as a newly formed entity.

“Participants unanimously reported that the “AI governance committee should be a subcommittee reporting up to DHC,” leveraging its members’ expertise and facilitating smoother implementation, rather than as a newly formed entity.”

May combine two claims.

Participants identified key elements essential to this governance structure, including clearly defined stakeholder roles and responsibilities, established monitoring processes with metrics and frequency, and specific criteria for decommissioning AI products.

“Participants identified several key elements essential to this governance structure, including clearly defined stakeholder roles and responsibilities, established monitoring processes with metrics and frequency, and specific criteria for decommissioning AI products.”

Probably not a substantial claim: may not pass S3 (load-bearing).

One additional vendor-developed AI product was undergoing local evaluation.

“with one additional vendor-developed product undergoing local evaluation.”

May combine two claims.

Participants appreciated the inclusion of a centralized inventory of algorithms and lifecycle management process, both of which are currently lacking within the organization.

“For example, participants appreciated the inclusion of a centralized inventory of algorithms and lifecycle management process, both of which are currently lacking within the organization.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants felt that the general AI governance framework could mirror existing non-AI technology governance.

“While participants felt that the general AI governance framework could mirror existing non-AI technology governance, they sought further clarification on elements unique to AI oversight.”

May combine two claims.

IT teams successfully extracted data from the EHR, shared it with vendors to facilitate model adaptation and evaluation under appropriate agreements, and reintegrated it into the EHR system.

“IT teams successfully extracted data from the EHR, shared it with vendors to facilitate model adaptation and evaluation under appropriate agreements, and reintegrated it into the EHR system.”

May combine two claims.

Effective practices such as feasibility assessment, clinician engagement in solution design, and establishment of communication channels were occasionally implemented but not consistently applied across projects.

“while certain best practices–such as feasibility assessment, clinician engagement in solution design, and the establishment of communication channels–were occasionally implemented, they were not consistently applied across projects.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants particularly appreciated the inclusion of ethics consultation through the ethics and legal subcommittee, which addressed the needs identified in the interviews.

“They particularly appreciated the inclusion of ethics consultation through the ethics and legal subcommittee, which addressed the needs identified in the interviews.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants stressed the importance of balancing structure and innovation to ensure that the governance process supports AI implementation without undue delays.

“Participants stressed the importance of balancing structure and innovation to ensure that the governance process supports AI implementation without undue delays.”

May combine two claims.

Team members were open to expanding their expertise in AI governance and demonstrated a commitment to adapt to the organization's evolving digital health needs.

“they were open to expanding their expertise in AI governance and demonstrated a commitment to adapt to the organization’s evolving digital health needs.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants suggested that before recruiting nominees, it is essential to establish and document expectations for AI governance committee members.

“Participants also suggested that before recruiting nominees, it is essential to establish and document expectations for AI governance committee members.”

Classified as probably not a substantial claim. May combine two claims.

Participants supported stratifying oversight levels based on the risk associated with each product, advocating for more rigorous examination of high-risk AI compared to low-risk AI.

“While participants supported stratifying oversight levels based on the risk associated with each product—advocating for more rigorous examination of high-risk AI compared to low-risk AI—developing the consensus around the purview of AI governance was found to be challenging.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants advocated for collaboration between AI ethics experts and computer scientists to identify potential risks and downstream impacts associated with AI adoption.

“Participants advocated for collaboration between AI ethics experts and computer scientists to identify potential risks and downstream impacts associated with AI adoption.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Additional key takeaways from the workshops included clarifying the scope of AI governance and defining the enforcement boundaries for the AI governance committee.

“Additional key takeaways from the workshops included clarifying the scope of AI governance and defining the enforcement boundaries for the AI governance committee.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants emphasized that all stakeholders, regardless of role, should possess basic technical knowledge-particularly related to EHRs.

“Participants emphasized that all stakeholders, regardless of role, should possess basic technical knowledge—particularly related to EHRs.”

Classified as probably not a substantial claim.

The importance of ethics consultation was highlighted, particularly in monitoring health equity dimensions.

“The importance of ethics consultation was also highlighted, particularly in monitoring health equity dimensions.”

Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing). May combine two claims.

Participants reported that this collaboration, bringing together clinical, technical, and business expertise, would ensure the selection of high-quality, appropriate data elements for AI models, thorough documentation of data usage processes, proper approvals for data utilization, and alignment with business objectives.

“They reported that this collaboration, bringing together clinical, technical, and business expertise, would ensure the selection of high-quality, appropriate data elements for AI models, thorough documentation of data usage processes, proper approvals for data utilization, and alignment with business objectives.”

May combine two claims.

The AI governance committee was structured under the leadership of a single executive and clinician leader and included subcommittees representing a diverse range of stakeholders.

“The AI governance committee was structured under the leadership of a single executive and clinician leader and included subcommittees representing a diverse range of stakeholders.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Participants brainstormed key measures of success for evaluating the effectiveness of AI governance and prioritized trust, user satisfaction, efficiency, compliance, and risk mitigation as foundational measures.

“They then brainstormed key measures of success for evaluating the effectiveness of AI governance and prioritized trust, user satisfaction, efficiency, compliance, and risk mitigation as foundational measures.”

May combine two claims.

Limited AI expertise among key stakeholders added another layer of difficulty, requiring efforts to ensure the policies were accessible and actionable to those responsible for implementation.

“Limited AI expertise among key stakeholders added another layer of difficulty, requiring efforts to ensure the policies were accessible and actionable to those responsible for implementation.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants proposed collaboration between a clinical systems and informatics team and a business team.

“For instance, participants proposed collaboration between a clinical systems and informatics team and a business team.”

May combine two claims.

Workshop participants were able to formulate organizational policies around AI governance, which were subsequently formally adopted by the institution.

“As the ultimate outcome of this study, workshop participants were able to formulate organizational policies around AI governance, which were subsequently formally adopted by the institution.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants noted the absence of an explicit operational plan for covering the ongoing costs of technical capabilities and infrastructures required to support the AI lifecycle.

“Regarding costs, participants noted the absence of an explicit operational plan for covering the ongoing costs of technical capabilities and infrastructures required to support the AI lifecycle.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants recommended that these external teams take responsibility for owning and operationalizing the technical capabilities and infrastructure throughout the entire AI lifecycle.

“Additionally, participants recommended that these external teams take responsibility for owning and operationalizing the technical capabilities and infrastructure throughout the entire AI lifecycle.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants recognized the need for a dedicated budget to operationalize the AI governance system.

“They recognized the need for a dedicated budget to operationalize the AI governance system.”

May combine two claims.

The policies defined the people, process, technology, and operations needed for AI governance and were structured according to the organization's standard policy framework.

“Using the PPTO framework as a roadmap, the policies defined the people, process, technology, and operations needed for AI governance and were structured according to the organization’s standard policy framework.”

May combine two claims.

As of October 2024, the organization has approved the policies, and as a result, an AI governance committee was successfully established in alignment with the policy recommendations.

“As of October 2024, the organization has approved the policies, and as a result, an AI governance committee was successfully established in alignment with the policy recommendations.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants proposed disseminating the documentation and educational materials throughout the organization using existing standard processes and communication channels.

“They proposed disseminating them throughout the organization using existing standard processes and communication channels.”

Classified as probably not a substantial claim.

Participants agreed that even if an AI product complies with regulations and standards set by health authorities or regulatory bodies, if it directly interfaces with patients or informs patient care decisions, it should be fully governed.

“Even if an AI product complies with regulations and standards set by health authorities or regulatory bodies, if it directly interfaces with patients or informs patient care decisions, it should be fully governed.”

Classified as probably not a substantial claim.

Participants were reminded that the committee's role is to provide oversight-ensuring AI initiatives are ethical and responsible-rather than directly leading implementation.

“However, they were reminded that the committee’s role is to provide oversight—ensuring AI initiatives are ethical and responsible—rather than directly leading implementation.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants nominated teams outside the AI governance committee to take responsibility for finalizing the specifications of the technical capabilities and infrastructure.

“They nominated teams outside the AI governance committee to take responsibility for finalizing the specifications of the technical capabilities and infrastructure.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Participants debated whether AI products used for non-clinical purposes-such as clinical research, billing, and scheduling-as well as federally approved products, should fall under governance, with views varying considerably.

“For example, participants debated whether AI products used for non-clinical purposes—such as clinical research, billing, and scheduling—as well as federally approved products, should fall under governance. Views varied considerably.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Participants proposed engaging patient advisors on the AI governance committee, collaborating with the Patient Experience Team, and leveraging insights from the Patient Family Advisory Council to effectively integrate patient perspectives into AI development and address patient needs.

“To effectively integrate patient perspectives into AI development and address patient needs, participants proposed engaging patient advisors on the AI governance committee, collaborating with the Patient Experience Team, and leveraging insights from the Patient Family Advisory Council.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Participants indicated that cost coverage would likely depend on the specific AI use case and the availability of funding sources.

“Participants indicated that cost coverage would likely depend on the specific AI use case and the availability of funding sources.”

Classified as probably not a substantial claim.

Participants ultimately agreed that the risk level of an AI product is contingent upon the specific use case.

“Ultimately, participants agreed that the risk level of an AI product is contingent upon the specific use case.”

Classified as probably not a substantial claim.

The lack of uniformity highlighted the need for a more structured approach to ensure reliable and effective AI governance across the organization.

“This lack of uniformity highlighted the need for a more structured approach to ensure reliable and effective AI governance across the organization.”

May combine two claims.

The utility of risk assessment frameworks within healthcare delivery organizations remains limited because they were not specifically designed for the healthcare context and do not offer actionable guidance for establishing AI governance structures at the organizational level.

“However, their utility within healthcare delivery organizations remains limited, as they were not specifically designed for the healthcare context and do not offer actionable guidance for establishing AI governance structures at the organizational level.”

May combine two claims.

The Process domain was the most critical area to address, as it structures the AI governance system and underscores the need for a centralized, standardized governance process.

“The Process domain was the most critical area to address, as it structures the AI governance system and underscores the need for a centralized, standardized governance process, which was strongly emphasized in the interviews.”

Classified as probably not a substantial claim. May combine two claims.

Robust AI governance mitigates risks, safeguards patient safety, and supports the responsible integration of AI into healthcare.

“It mitigates risks, safeguards patient safety, and supports the responsible integration of AI into healthcare16.”

• cites [16] Reddy, S., Allan, S., Coghlan, S. & Cooper, P. A governance model for the application of AI in health care. J. Am. Med. Inform. Assoc. 26, 1034–1037 (2019).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

The study generates empirical evidence of how a conceptual AI governance framework can be implemented within healthcare delivery organizations to drive organizational change, addressing a critical gap in the field.

“It also addresses a critical gap in the field by generating empirical evidence of how a conceptual AI governance framework can be implemented within healthcare delivery organizations to drive organizational change.”

• AssumptionRests on an assumption: Few studies offer practical insights into the process of establishing AI governance within healthcare organizations.

Probably not a substantial claim: may not pass S3 (load-bearing).

The Technology domain of the PPTO framework details the infrastructure and technical capabilities necessary to oversee AI tools effectively throughout their lifecycles.

“The Technology domain details the infrastructure and technical capabilities necessary to oversee AI tools effectively throughout their lifecycles.”

• cites [35] Kim, J. Y., Hasan, A., Balu, S. & Sendak, M. People, process, technology, and operations (PPTO) framework for organizational AI governance in healthcare. npj Digit. Med.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

The People domain of the PPTO framework specifies the personnel needed for AI governance, including the structure of the AI governance committee, required areas of expertise, defined roles and responsibilities, and strategies for managing membership over time.

“The People domain specifies the personnel needed for AI governance, outlining the structure of the AI governance committee, required areas of expertise, defined roles and responsibilities, and strategies for managing membership over time.”

• cites [35] Kim, J. Y., Hasan, A., Balu, S. & Sendak, M. People, process, technology, and operations (PPTO) framework for organizational AI governance in healthcare. npj Digit. Med.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S4 (based).

The PPTO framework enables a systematic and comprehensive assessment of the capabilities and resources required to establish AI governance.

“It enables a systematic and comprehensive assessment of the capabilities and resources required to establish AI governance.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The Operations domain of the PPTO framework describes the organizational capabilities required to operationalize AI governance, including executive sponsorship, accountability for committee activities, budget planning, and metrics to evaluate governance effectiveness.

“The Operations domain describes the organizational capabilities required to operationalize AI governance, including executive sponsorship, accountability for committee activities, budget planning, and metrics to evaluate governance effectiveness.”

• cites [35] Kim, J. Y., Hasan, A., Balu, S. & Sendak, M. People, process, technology, and operations (PPTO) framework for organizational AI governance in healthcare. npj Digit. Med.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

The Process domain of the PPTO framework outlines a governance process that balances innovation with risk, detailing key decision points across the AI lifecycle and the associated documentation required.

“The Process domain outlines a governance process that balances innovation with risk, detailing key decision points across the AI lifecycle and the associated documentation required.”

• cites [35] Kim, J. Y., Hasan, A., Balu, S. & Sendak, M. People, process, technology, and operations (PPTO) framework for organizational AI governance in healthcare. npj Digit. Med.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S2 (checkable).

Robust AI governance ensures that AI solutions are transparent, equitable, and aligned with ethical and regulatory standards.

“Robust AI governance ensures that AI solutions are transparent, equitable, and aligned with ethical and regulatory standards.”

• cites [16] Reddy, S., Allan, S., Coghlan, S. & Cooper, P. A governance model for the application of AI in health care. J. Am. Med. Inform. Assoc. 26, 1034–1037 (2019).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The AI Risk Management Framework (AI RMF) was developed by the U.S. National Institute of Standards and Technology (NIST), and the Algorithmic Impact Assessment (AIA) tool was introduced by the Government of Canada.

“For example, the AI Risk Management Framework (AI RMF) was developed by the U.S. National Institute of Standards and Technology (NIST)28, and the Algorithmic Impact Assessment (AIA) tool was introduced by the Government of Canada29.”

• cites [28] National Institute of Standards and Technology (NIST). AI Risk Management Framework. https://www.nist.gov/itl/ai-risk-management-framework (accessed 2025).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [29] Government of Canada. Algorithmic Impact Assessment tool. https://www.canada.ca/en/government/system/digital-government/digital-government-innovations/responsible-use-ai/algorithmic-impact-assessment.html (accessed 2025).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S4 (based).

The PPTO framework was selected for its practicality, comprehensiveness, and relevance within healthcare.

“The PPTO framework was selected for its practicality, comprehensiveness, and relevance within healthcare.”

Probably not a substantial claim: may not pass S4 (based).

By addressing AI technology and the people, process, and operations that support it, the PPTO framework ensures a holistic approach to managing the complexities of AI governance.

“By addressing not only AI technology but also the people, process, and operations that support it, the framework ensures a holistic approach to managing the complexities of AI governance.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

The TOE framework's broad, high-level guidance lacks the specificity necessary to support concrete, actionable steps for implementing governance structures in healthcare settings.

“Its broad, high-level guidance—intended for general applicability across fields—lacks the specificity necessary to support concrete, actionable steps for implementing governance structures in healthcare settings.”

Classified as probably not a substantial claim.

The Technology-Organization-Environment (TOE) framework explains how organizations adopt new technologies by considering internal factors such as technological capabilities and organizational resources, and external factors including the availability of technologies in the market and the regulatory environment.

“The Technology-Organization-Environment (TOE) framework explains how organizations adopt new technologies by considering both internal factors—such as technological capabilities and organizational resources—and external factors, including the availability of technologies in the market and the regulatory environment22.”

• cites [22] Tornatzky, L. G. & Fleischer, M. The Processes of Technological Innovation (Lexington Books, 1990).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing). May combine two claims.

This approach also aims to establish a transparent system for sharing information with patients and communities, reflecting their needs in AI governance.

“This approach also aims to establish a transparent system for sharing information with patients and communities, reflecting their needs in AI governance.”

Probably not a substantial claim: may not pass S2 (checkable).

AI governance requires shared decision-making and open dialog, especially in settings where AI knowledge is still emerging and where organizational hierarchies may constrain participation.

“It requires shared decision-making and open dialog, especially in settings where AI knowledge is still emerging and where organizational hierarchies may constrain participation.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

Such engagement not only ensures the governance system is aligned with organizational realities but also fosters broader organizational buy-in.

“Such engagement not only ensures the governance system is aligned with organizational realities but also fosters broader organizational buy-in.”

Probably not a substantial claim: may not pass S2 (checkable).

Building AI governance capacity requires investments in education on AI in healthcare.

“Lastly, building AI governance capacity requires investments in education on AI in healthcare.”

Probably not a substantial claim: may not pass S3 (load-bearing).

To mitigate potential bias, the study ensured diverse representation across clinical, technical, and operational domains, organizational levels, and years of experience.

“To mitigate potential bias, we ensured diverse representation across clinical, technical, and operational domains, organizational levels, and years of experience.”

May combine two claims.

The study was considered a quality improvement project/initiative that did not involve the collection of human subjects data (or identifiable data).

“This study was considered a quality improvement project without the involvement of data collection from human subjects.”

• Also stated in: § Part 1: Stakeholder interviews · plain

  “This study was considered a quality improvement initiative and did not involve the collection of identifiable data about human subjects.”

Probably not a substantial claim: may not pass S2 (checkable).

The development of AI governance is inherently collaborative.

“Third, the development of AI governance is inherently collaborative.”

Probably not a substantial claim: may not pass S2 (checkable).

Creating safe spaces for engagement, and valuing the perspectives of all stakeholders, enables organizations to build more thoughtful, inclusive, and responsive governance systems.

“Creating safe spaces for engagement, and valuing the perspectives of all stakeholders, enables organizations to build more thoughtful, inclusive, and responsive governance systems.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

These methods provide structured yet flexible formats for sharing, gathering, and synthesizing diverse insights.

“These methods provide structured yet flexible formats for sharing, gathering, and synthesizing diverse insights.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Without clear boundaries, the committee risked becoming overwhelmed and inefficient.

“Without clear boundaries, the committee risked becoming overwhelmed and inefficient.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

By incorporating diverse stakeholder voices, governance systems are more likely to be effective, relevant, and ethically grounded.

“By incorporating diverse stakeholder voices, governance systems are more likely to be effective, relevant, and ethically grounded.”

May combine two claims.

This research supports and extends the PPTO framework by validating its practical usability and relevance, while also demonstrating its adaptability across diverse organizational contexts and varying levels of AI maturity.

“This research supports and extends the PPTO framework by validating its practical usability and relevance, while also demonstrating its adaptability across diverse organizational contexts and varying levels of AI maturity.”

Probably not a substantial claim: may not pass S3 (load-bearing).

All participants provided verbal consent/informed consent to participate and to have anonymized data used in qualitative analyses.

“All participants provided verbal consent to participate in the study and to have anonymized data used in qualitative analyses.”

• Also stated in: § Part 1: Stakeholder interviews · plain

  “All participants provided verbal informed consent to participate in interviews, to have the interviews recorded, and to allow anonymized data to be used in qualitative analysis.”

• Also stated in: § Part 2: Co-design workshops · plain

  “All participants provided verbal informed consent to participate in the workshops and to allow anonymized data to be used in qualitative analysis.”

• AssumptionRests on an assumption: Verbal consent is sufficient for participation in this quality improvement study.

May combine two claims.

Ethics board approval was not required for this study (it was exempt from IRB approval).

“As such, ethics board approval was not required.”

• Also stated in: § Part 1: Stakeholder interviews · plain

  “Thus, it was exempt from IRB approval.”

• AssumptionRests on an assumption: The study meets the regulatory criteria for exemption from ethics board approval as a quality improvement project.

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

Education about AI's technical underpinnings, clinical applications, risks, and ethical considerations equips stakeholders to participate meaningfully in governance efforts and fosters long-term accountability, transparency, and trust.

“Education about AI’s technical underpinnings, clinical applications, risks, and ethical considerations equips stakeholders to participate meaningfully in governance efforts and fosters long-term accountability, transparency, and trust.”