By Dr. Freddie Seba © 2026 Freddie Seba. All rights reserved.
AMIA — American Medical Informatics Association: What the clinical informatics governance framework covers — and what it leaves to you
Applying the AI Minimum Viable Governance (AI-MVG) framework and the Seba 12 Ps of Responsible AI
Fifth installment. Previously: Anthropic, OpenAI, Google Gemini, and NIST CAISI. Each stands alone — but the pattern across all five is where the real governance story begins.
This miniseries examines frameworks shaping AI governance globally and surfaces what leaders need to know about the gaps, whether intentional or not. Not to critique AI Labs or organizations. But to ask three questions every institution should be asking before deployment, not after a patient outcome forces the conversation.
This week: AMIA — the American Medical Informatics Association.
Question One — What was the framework designed to govern?
AMIA is the professional home of clinical informatics in the United States and one of the most consequential voices shaping the governance of AI in healthcare settings. Its governance work is substantive, sustained, and technically grounded across three decades of health informatics practice.
AMIA’s framework established six core stewardship principles for health data: accountability, chain of trust, transparency, oversight, data quality, and individual participation. These principles represent the foundation of responsible health data governance in the United States — peer-reviewed, practically grounded, and widely adopted across health systems, federal agencies, and research institutions.
AMIA’s position on AI in clinical decision support is that its use is neither inherently good nor bad — it is the governance surrounding deployment that determines outcomes. AMIA has established an AI Task Force to produce position statements, white papers, and policy advocacy, and actively collaborates with CHAI, the National Academy of Medicine, and the Health AI Partnership to translate principles into institutional practice. AMIA also convenes federal policymakers, clinical informaticists, and patient representatives for structured dialogue on human oversight in clinical AI — including a 2026 policy forum dedicated to human judgment in the age of clinical AI.
It governs the standards, ethics, and practice of clinical informatics. Seriously. At depth. With three decades of stewardship behind it.
Table 1 — Framework snapshot
| Dimension | AMIA coverage | Institutional gap |
| Purpose | Ethical, evidence-based clinical AI deployment | Institutional purpose alignment not evaluated |
| People | Clinician and informaticist professional standards | Institution must assign named accountability |
| Processes | Stewardship framework and CDS governance guidance | Institutional deployment process not governed |
| Policy | Federal advocacy and position statements | No institutional change-response process |
| Privacy | HIPAA alignment; health data stewardship principles | Institutional data flows beyond HIPAA not mapped |
| Performance | Clinical performance standards for CDS | Institutional AI performance monitoring not set |
| Procurement | Not addressed | Vendor dependency and exit terms not mapped |
| Partnerships | CHAI, NAM, Health AI Partnership collaboration | Institutional third-party clinical AI risk not mapped |
| Predictability | Not addressed | No institutional continuity plan for clinical AI changes |
| Protection | Patient safety principles; HHS Section 1557 alignment | Institution must implement its own clinical protections |
| Proof | Peer-reviewed frameworks and position papers | Institutional audit trail for clinical AI rarely established |
| Precedent | Not assigned | Who owns decisions when clinical AI produces unexpected outcomes? |
Question Two — The Governance Gap: what falls outside the frame?
Institutional deployment accountability.
AMIA’s framework establishes the principles that should govern clinical AI — accountability, chain of trust, transparency, oversight, data quality, and individual participation. These principles are serious, peer-reviewed, and practically grounded in three decades of health data stewardship work.
What the framework was not designed to do is operationalize those principles inside your institution. AMIA calls for the creation of new bodies, groups, or departments that govern implementation and use of AI within an institution. That recommendation is important and forward-looking. Translating it into named roles, defined decision ownership, and institutional accountability structures is work that belongs to the institution — not to AMIA.
Consider the questions every health system deploying clinical AI should be able to answer:
- Who is accountable for each AI-assisted clinical decision in your health system?
- Who reviews vendor model updates before they affect clinical workflows?
- Who owns the institutional response when a clinical AI tool underperforms or produces unexpected outputs?
- Who determines whether a vendor’s updated model still meets the clinical and ethical thresholds your institution set at the time of deployment?
AMIA provides the principles and the professional standards to frame those questions. Only the institution can provide the answers — and building the governance structures to answer them is precisely what AI Minimum Viable Governance was designed to support.
Table 2 — Governance gap: AMIA vs. AI-MVG
| Dimension | AMIA’s framework | What AI-MVG requires |
| Stewardship principles | Established — accountability, chain of trust, transparency | Must be operationalized institutionally |
| Clinical decision ownership | Calls for institutional governance bodies | Institution must create, name, and empower them |
| Adverse outcome accountability | Principles established; institutional assignment not specified | Named executive accountability required |
| Vendor dependency | Not addressed | Must be mapped before deployment |
| Change management | Not addressed | Triggers must be defined before deployment |
| Primary question | “Is clinical AI being used ethically?” | “Who owns the decision and its consequences?” |
Question Three — What governance must institutions build before deployment?
This is where AI Minimum Viable Governance (AI-MVG) begins. Not after an adverse event triggers a root cause analysis — before deployment, when the institution still has the leverage to govern with intention.
- Vendor dependency mapping — know what breaks if the clinical AI vendor changes its model, terms, or training data.
- Clinical decision ownership — name which clinician, role, or committee is accountable for each AI-assisted decision category
- Policy monitoring — establish a process to track vendor and regulatory changes in real time.
- Change management triggers — define which levels of model updates, term changes, or performance drift require institutional review.
- Contingency planning — document what happens if a clinical AI tool is suspended, recalled, or found to produce unexpected outputs.
- Executive accountability — assign a named leader responsible for clinical AI governance outcomes, not just adoption metrics
This week’s 12 Ps lens: Precedent
Most health systems have deployed clinical AI. Very few have answered the precedent question: when the system produces an unexpected outcome, who owns that decision?
AMIA’s framework governs the ethics and standards of clinical informatics practice. It does not — and was never designed to — assign precedent inside your institution. Precedent — the twelfth P — means establishing in advance who owns accountability for AI-assisted clinical decisions, what the escalation path is when outcomes are unexpected, and how the institution will respond when a vendor’s model requires review.
The governance structures that answer those questions are not a critique of clinical AI. They are what makes clinical AI trustworthy enough to scale.
Table 3 — Full 12 Ps lens: AMIA governance coverage
| Dimension | AMIA coverage | Institutional gap |
| Purpose | Ethical, evidence-based clinical AI deployment | Institutional purpose alignment not evaluated |
| People | Clinician and informaticist professional standards | Institution must assign named accountability |
| Processes | Stewardship framework and CDS governance guidance | Institutional deployment process not governed |
| Policy | Federal advocacy and position statements | No institutional change-response process |
| Privacy | HIPAA alignment; health data stewardship principles | Institutional data flows beyond HIPAA not mapped |
| Performance | Clinical performance standards for CDS | Institutional AI performance monitoring not set |
| Procurement | Not addressed | Vendor dependency and exit terms not mapped |
| Partnerships | CHAI, NAM, Health AI Partnership collaboration | Institutional third-party clinical AI risk not mapped |
| Predictability | Not addressed | No institutional continuity plan for clinical AI changes |
| Protection | Patient safety principles; HHS Section 1557 alignment | Institution must implement its own clinical protections |
| Proof | Peer-reviewed frameworks and position papers | Institutional audit trail for clinical AI rarely established |
| Precedent | Not assigned | Who owns decisions when clinical AI produces unexpected outcomes? |
AMIA governs the ethics of clinical informatics. Institutional governance governs the consequences of clinical AI deployment. One governs the standard of practice. The other governs the decision. That is the foundation of AI Minimum Viable Governance — and in healthcare, it is the difference between a governance principle and a governance practice.
For executive briefings, board workshops, and keynote presentations → freddieseba.com
About the Author
Dr. Freddie Seba helps boards, trustees, and executive leadership teams build practical AI governance before AI failures make governance unavoidable. Scholar-operator, Silicon Valley founder, and global executive — EdD, USF · MBA, Yale · MA, Stanford.
This analysis is part of Ungoverned: Applied Frameworks Under the Lens, a recurring miniseries that applies the AI-MVG framework and the Seba 12 Ps of Responsible AI to organizations shaping AI governance globally — including AI labs, governments, standards bodies, and international actors such as the EU, China, UNESCO, and Singapore. Drafted with AI-assisted tools. Final editorial judgment and responsibility remain with the author.
Sources
AMIA Current Policy Priorities — amia.org/public-policy/current-policy-priorities
AMIA Position Paper on Adaptive CDS: Policy Framework for AI/ML-Driven Decision Support — amia.org/news-publications/amia-position-paper-details-policy-framework-aiml-driven-decision-support
AMIA AI Task Force Call for Volunteers — amia.org/news-publications/call-volunteers-artificial-intelligence-task-force
AMIA 2026 Policy Forum: Human Judgment in the Age of Clinical AI — amia.org/education-events/Human-Judgement-in-the-Age-of-Clinical-AI
JMIR: From Data Stewardship to Model Stewardship (June 2026) — jmir.org/2026/1/e97859
PMC: From Data Stewardship to Model Stewardship — pmc.ncbi.nlm.nih.gov/articles/PMC13240980
Drafted with AI-assisted tools. Final editorial judgment and responsibility remain with the author.
#AIGovernance #ResponsibleAI #AIMVG #AILeadership #Ungoverned #HealthcareAI #ClinicalInformatics #BoardGovernance #AIPolicy #PatientSafety #CMIO #CIO
Leave a Reply