Core challenges
The value of AI lies not only in the performance of an individual model, but also in whether it can address real healthcare needs. Interdisciplinary teams need a shared language, comparable data and evaluation methods, appropriate governance, and the ability to translate research into practical improvements.
TSMAI plans to promote collaboration through research exchange, interdisciplinary education, shared specifications, application evaluation, and responsible translation of research into practice. Multicenter validation and care processes for Clinical AI are only part of this scope; health promotion, healthcare quality, management, and administrative applications are also among the directions for work. This scope does not indicate that government authorization has been obtained or formal partnerships have been established.
Scope and mission
The mission is to promote research, education, governance, and safe applications of Medical AI, with the aim of improving healthcare quality, patient safety, health promotion, and interdisciplinary collaboration. Medical AI includes clinical care as well as medical data science, healthcare management, talent development, and translation of innovation into practice.
Five keywords for the mission: Education, Standards, Evidence, Governance, Research translation.
DRAFT revised wording for the charter’s mission statement — based on the discussion draft provided on 2026-09-06 and the Medical AI scope, awaiting review and approval through the formal charter process; the official Chinese name 台灣醫療人工智慧學會, the English name Taiwan Society of Medical Artificial Intelligence, and the abbreviation TSMAI have been confirmed by the person responsible for preparation.
Directions for charter revision — DRAFT
This summary draws on Articles 3 and 6 and the organizational structure in the charter discussion draft to broaden the Society’s scope of work from Clinical AI to Medical AI; it is neither the complete charter nor rules already in effect.
- Research and exchange: Medical AI, smart healthcare, and medical data science; shared data specifications, evaluation metrics, multicenter collaboration, and tracking of results.
- Applications and improvement: clinical care, health promotion, healthcare quality, patient safety, healthcare processes, and healthcare management; design fit-for-purpose evaluation and retain human oversight.
- Education and talent: courses, seminars, workshops, publications, and domestic and international academic exchange; encourage participation by physicians, healthcare professionals, researchers, engineers, and students.
- Governance and public dialogue: data governance, information security, privacy, ethics, model validation, implementation evaluation, monitoring, auditing, and the allocation of responsibility; develop policy and practice recommendations.
- Collaboration and research translation: connect healthcare institutions, academic and research institutions, and industry to advance research, education, outreach, evaluation, and exchange related to the mission; preserve publication rights and academic independence.
The numbers of directors and supervisors, their terms of office, and the division of responsibilities for the first term in the original draft are proposed options; membership eligibility, membership fees, and organizational rules remain to be confirmed and are not being finalized through the website. Confirmation of the name does not amount to registration as a legal entity or approval of the charter in its entirety.
Three North Star goals for the existing clinical engineering pathway
The clinical engineering goals of the v2.0 blueprint are retained below; new projects, resources, and schedules for the broader Medical AI scope require separate review and approval.
- Within one year: Develop an AI workflow that makes a noticeable difference to physicians
- Within three years: Develop a clinical AI platform that can undergo multicenter validation and seek regulatory authorization
- Within five years: Become a center for Medical AI standards and evidence in Taiwan
Five-year timeline plan
The following summarizes the existing clinical engineering pathway in the five-year strategic blueprint v2.0; it does not represent a plan for the entire Medical AI field. Each phase remains a direction for planning rather than an established fact; refer to each evidence page for maturity.
-
Phase 0 (0–6 months): Preparations and setting direction
Plans include preparing the charter and preparing to establish an incorporated association, recruiting core hospitals and core physicians, and planning the committee structure; identifying the first three demonstration projects (pre-visit SOAP, heart failure GDMT, and AF anticoagulation checks); and developing the PHI scrubbing foundation and audit ledger.
-
Phase 1 (Y1): Making a noticeable difference to physicians + a minimum viable platform
Plans include deploying core systems at the founding hospital sites; starting Level 1 educational courses; testing pre-visit SOAP in outpatient clinics, conducting multicenter retrospective validation of heart failure GDMT, and launching AF anticoagulation checks; submitting the first papers and holding the first annual meeting.
-
Phase 2 (Y2–Y3): Multicenter expansion, regulatory authorization, and standardization
Plans include formalizing the federated evidence platform, initiating prospective silent trials, developing a Taiwan-specific Model Card standard v1.0 and a medical MCP repository v0.1, and beginning formal operation of Clinical AI Studio; the aim is to complete the first submission or formal consultation for SaMD regulatory authorization.
-
Phase 3 (Y4–Y5): A standards and evidence center
Plans include voluntary "clinical AI deployment certification" (certifying processes, not models), a national clinical AI registry, publication of the Compendium of Clinical AI Standards for Taiwan v1.0, methodology for evidence dossiers for national health insurance reimbursement, and international dissemination.
Organizational structure — DRAFT
Based on the newly provided discussion draft, the planned structure consists of the General Assembly, Board of Directors, Board of Supervisors, Secretariat, and seven specialized committees. The following outlines proposed responsibilities; it does not indicate that the organization or committees have been formally established. Personnel and detailed rules are pending review and approval.
| Committee | Proposed responsibilities |
|---|---|
| Committee on Medical AI Applications and Multicenter Research (proposed name expansion) | Interdisciplinary Medical AI research, shared data specifications, evaluation methods, and collaboration across institutions; the name in the original draft is "Committee on Clinical Applications and Multicenter Research" |
| Committee on Data Governance, Information Security, and Ethics | Data use, privacy, information security, bias assessment, auditing, and post-deployment monitoring |
| Committee on Education, Training, and Talent Development | Medical AI literacy, interdisciplinary courses, workshops, and talent development |
| Committee on Healthcare Quality and Patient Safety | Outcome evaluation, workflow burden, quality indicators, and patient safety |
| Committee on Industry–Academia Collaboration and Translation of Innovation into Practice | Translation of needs into practice, collaboration evaluation, research integrity, and conflict of interest management |
| Committee on Policy, Regulation, and External Affairs | Policy research, ethical and regulatory issues, external exchange, and public communication |
| Committee on Digital Tools and Member Services | Website, knowledge and project management, member services, and secure administrative digital workflows |
Changes to the number and names of committees do not remove the role separation requirements of D-03: the research network may conduct certification under an approval framework, but the certification committee and reviewer must remain separate from development, consulting, sponsored research, and validation activities for the same case, as well as the project under review; Studio must not certify its own work. See process-based independence firewall (Mandarin only).
Committee membership list — in preparation. The list will be published once reviewed, approved, and authorized.
Charter and governance documents
See the DRAFT governance and responsibility entry point (Mandarin only), role separation and COI (Mandarin only), Taiwan’s Food and Drug Administration (TFDA) and institutional review board (IRB) boundaries (Mandarin only), and safeguards for patients / the public (Mandarin only).
The following documents will be made available on this page once finalized:
- Society charter — DRAFT / pending review and approval. A summary of the directions for revision is currently available; the complete approved version is not yet available.
- Conflict of interest (COI) policy — in preparation
- Data governance charter — in preparation