eduKateSG · Why Science?
Read an AI device list as a set of regulatory decisions
Use FDA’s current AI-Enabled Medical Device List through submission numbers, pathways, decision dates, specialties, counts, labels, software versions, PCCPs and postmarket records.
Artificial intelligence now appears in devices across imaging, monitoring, diagnosis and treatment. FDA’s public AI-Enabled Medical Device List gives students a visible map of products the agency has identified as AI-enabled and authorized for marketing.
The list is most useful as a doorway. Follow each submission number into its 510(k), De Novo or PMA record, then read the intended use, decision summary and current labeling before making a claim.
Did you know? FDA reported more than 1,600 authorised AI-enabled medical devices as of September 2026. That is a dated total, and it counts regulatory list entries rather than proving how many distinct algorithms are in clinical use.
This guide supports digital-health and regulatory literacy. It does not compare products or recommend medical care.
Chapter index
Chapters 1–9: list fields, identifiers, pathways and dated counts
- 1. The list identifies marketed devices
- 2. Authorization comes from established pathways
- 3. The submission number is the evidence key
- 4. Decision date fixes the regulatory moment
- 5. Device name and company need normalisation
- 6. Panel or specialty supports grouping
- 7. AI-enabled is a transparency label
- 8. The list is not comprehensive forever
- 9. The September 2026 total is dated
Chapters 10–17: labeling, software versions, changes and postmarket evidence
- 10. A count needs a defined unit
- 11. Radiology concentration needs context
- 12. Authorization is not comparative effectiveness
- 13. Labeling defines the intended use
- 14. Software versions bound conclusions
- 15. PCCPs may govern some later changes
- 16. Postmarket evidence answers later questions
- 17. A strong list receipt follows the links
An AI-enabled medical-device list entry becomes useful when the device, submission number, decision date, specialty and source record remain connected. FDA says the AI-Enabled Medical Device List is intended to identify AI-enabled medical devices authorized for marketing in the United States. This turns an impressive total into an auditable set of decisions.
Try this
Evidence question: Which official page, guidance or decision supports the point about the list identifies marketed devices? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for the list identifies marketed devices. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Keep the identifier, retrieval date and source page in the note so the route can be repeated.
Evidence checkpoint
Record the evidence object for the list identifies marketed devices, its identifier or version context, and one limit that must remain visible.
FDA describes the list as a transparency resource for AI-enabled devices authorized for marketing in the United States. Entries connect to decisions such as 510(k) clearance, De Novo classification or PMA approval rather than an AI-specific marketing pathway. It also separates AI identification from proof of comparative clinical benefit.
Try this
Evidence question: Which official page, guidance or decision supports the point about authorization comes from established pathways? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for authorization comes from established pathways. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. If a field is missing, write “not shown” and follow the official documentation rather than inventing a value.
Evidence checkpoint
Record the evidence object for authorization comes from established pathways, its identifier or version context, and one limit that must remain visible.
A list entry reflects an authorization through an established pathway, not an endorsement of every use or model update. A K, DEN or PMA number links the list row to the official decision record, summary and applicant information. Submission numbers and decision summaries make the route reproducible.
Try this
Evidence question: Which official page, guidance or decision supports the point about the submission number is the evidence key? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for the submission number is the evidence key. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Before comparing two records, check that their object types, methods, versions and units really align.
Evidence checkpoint
Record the evidence object for the submission number is the evidence key, its identifier or version context, and one limit that must remain visible.
Counts and technology labels are dated because the list and FDA identification methods evolve. The date identifies when FDA made the marketing decision and prevents later list updates from being mistaken for the authorization date. Current labeling and software versions still bound any product claim.
Try this
Evidence question: Which official page, guidance or decision supports the point about decision date fixes the regulatory moment? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for decision date fixes the regulatory moment. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Finish by naming one supported statement and one stronger statement that still needs another evidence layer.
Evidence checkpoint
Record the evidence object for decision date fixes the regulatory moment, its identifier or version context, and one limit that must remain visible.
An AI-enabled medical-device list entry becomes useful when the device, submission number, decision date, specialty and source record remain connected. Trade names, applicant names and corporate changes can vary, so identifiers are safer than name-only matching. This turns an impressive total into an auditable set of decisions.
Try this
Evidence question: Which official page, guidance or decision supports the point about device name and company need normalisation? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for device name and company need normalisation. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Keep the identifier, retrieval date and source page in the note so the route can be repeated.
Evidence checkpoint
Record the evidence object for device name and company need normalisation, its identifier or version context, and one limit that must remain visible.
FDA describes the list as a transparency resource for AI-enabled devices authorized for marketing in the United States. FDA’s medical specialty field helps readers see where AI-enabled devices are concentrated without proving clinical performance. It also separates AI identification from proof of comparative clinical benefit.
Try this
Evidence question: Which official page, guidance or decision supports the point about panel or specialty supports grouping? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for panel or specialty supports grouping. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. If a field is missing, write “not shown” and follow the official documentation rather than inventing a value.
Evidence checkpoint
Record the evidence object for panel or specialty supports grouping, its identifier or version context, and one limit that must remain visible.
A list entry reflects an authorization through an established pathway, not an endorsement of every use or model update. Inclusion indicates FDA identified AI-related functionality; it does not mean every function of the device uses AI. Submission numbers and decision summaries make the route reproducible.
Try this
Evidence question: Which official page, guidance or decision supports the point about ai-enabled is a transparency label? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for ai-enabled is a transparency label. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Before comparing two records, check that their object types, methods, versions and units really align.
Evidence checkpoint
Record the evidence object for ai-enabled is a transparency label, its identifier or version context, and one limit that must remain visible.
Counts and technology labels are dated because the list and FDA identification methods evolve. FDA updates the resource as information and identification methods evolve, so an access date and downloaded snapshot matter. Current labeling and software versions still bound any product claim.
Try this
Evidence question: Which official page, guidance or decision supports the point about the list is not comprehensive forever? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for the list is not comprehensive forever. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Finish by naming one supported statement and one stronger statement that still needs another evidence layer.
Evidence checkpoint
Record the evidence object for the list is not comprehensive forever, its identifier or version context, and one limit that must remain visible.
An AI-enabled medical-device list entry becomes useful when the device, submission number, decision date, specialty and source record remain connected. FDA reported more than 1,600 authorized AI-enabled devices as of September 2026; this should not be repeated as a timeless count. This turns an impressive total into an auditable set of decisions.
Try this
Evidence question: Which official page, guidance or decision supports the point about the september 2026 total is dated? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for the september 2026 total is dated. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Keep the identifier, retrieval date and source page in the note so the route can be repeated.
Evidence checkpoint
Record the evidence object for the september 2026 total is dated, its identifier or version context, and one limit that must remain visible.
FDA describes the list as a transparency resource for AI-enabled devices authorized for marketing in the United States. Rows, submission numbers, products, models and software functions can produce different totals and should not be mixed. It also separates AI identification from proof of comparative clinical benefit.
Try this
Evidence question: Which official page, guidance or decision supports the point about a count needs a defined unit? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for a count needs a defined unit. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. If a field is missing, write “not shown” and follow the official documentation rather than inventing a value.
Evidence checkpoint
Record the evidence object for a count needs a defined unit, its identifier or version context, and one limit that must remain visible.
A list entry reflects an authorization through an established pathway, not an endorsement of every use or model update. A high share in one specialty can reflect data availability, workflow and classification patterns rather than superiority of AI in that field. Submission numbers and decision summaries make the route reproducible.
Try this
Evidence question: Which official page, guidance or decision supports the point about radiology concentration needs context? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for radiology concentration needs context. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Before comparing two records, check that their object types, methods, versions and units really align.
Evidence checkpoint
Record the evidence object for radiology concentration needs context, its identifier or version context, and one limit that must remain visible.
CHAPTER 12 OF 17 · Bound the AI claim
12. Authorization is not comparative effectiveness
Counts and technology labels are dated because the list and FDA identification methods evolve. A marketing decision does not by itself show that an AI-enabled device outperforms every clinician, device or standard of care. Current labeling and software versions still bound any product claim.
Try this
Evidence question: Which official page, guidance or decision supports the point about authorization is not comparative effectiveness? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for authorization is not comparative effectiveness. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Finish by naming one supported statement and one stronger statement that still needs another evidence layer.
Evidence checkpoint
Record the evidence object for authorization is not comparative effectiveness, its identifier or version context, and one limit that must remain visible.
An AI-enabled medical-device list entry becomes useful when the device, submission number, decision date, specialty and source record remain connected. The authorised indication, user, population, setting, inputs and outputs are found in the decision and labeling, not inferred from the phrase artificial intelligence. This turns an impressive total into an auditable set of decisions.
Try this
Evidence question: Which official page, guidance or decision supports the point about labeling defines the intended use? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for labeling defines the intended use. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Keep the identifier, retrieval date and source page in the note so the route can be repeated.
Evidence checkpoint
Record the evidence object for labeling defines the intended use, its identifier or version context, and one limit that must remain visible.
FDA describes the list as a transparency resource for AI-enabled devices authorized for marketing in the United States. Model architecture, training data, thresholds and deployment version can change, so product-level claims need exact software context. It also separates AI identification from proof of comparative clinical benefit.
Try this
Evidence question: Which official page, guidance or decision supports the point about software versions bound conclusions? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for software versions bound conclusions. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. If a field is missing, write “not shown” and follow the official documentation rather than inventing a value.
Evidence checkpoint
Record the evidence object for software versions bound conclusions, its identifier or version context, and one limit that must remain visible.
A list entry reflects an authorization through an established pathway, not an endorsement of every use or model update. An authorised Predetermined Change Control Plan can describe certain planned AI software modifications and controls. Submission numbers and decision summaries make the route reproducible.
Try this
Evidence question: Which official page, guidance or decision supports the point about pccps may govern some later changes? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for pccps may govern some later changes. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Before comparing two records, check that their object types, methods, versions and units really align.
Evidence checkpoint
Record the evidence object for pccps may govern some later changes, its identifier or version context, and one limit that must remain visible.
CHAPTER 16 OF 17 · Track the lifecycle
16. Postmarket evidence answers later questions
Counts and technology labels are dated because the list and FDA identification methods evolve. Recalls, safety communications, adverse-event reports and software updates belong to separate records with their own limitations. Current labeling and software versions still bound any product claim.
Try this
Evidence question: Which official page, guidance or decision supports the point about postmarket evidence answers later questions? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for postmarket evidence answers later questions. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Finish by naming one supported statement and one stronger statement that still needs another evidence layer.
Evidence checkpoint
Record the evidence object for postmarket evidence answers later questions, its identifier or version context, and one limit that must remain visible.
An AI-enabled medical-device list entry becomes useful when the device, submission number, decision date, specialty and source record remain connected. Record the row, submission number, date, pathway, device, company, specialty, decision summary, labeling, software version and access date. This turns an impressive total into an auditable set of decisions.
Try this
Evidence question: Which official page, guidance or decision supports the point about a strong list receipt follows the links? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the identifier, date, version and one limit for a strong list receipt follows the links. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Keep the identifier, retrieval date and source page in the note so the route can be repeated.
Evidence checkpoint
Record the evidence object for a strong list receipt follows the links, its identifier or version context, and one limit that must remain visible.
Contents · Previous chapter · Continue at the Science Learning Hub
An AI-device claim becomes reliable when the list row and decision record stay linked
The list improves transparency, while the linked authorization explains what FDA actually reviewed and decided. That distinction keeps an exciting technology label inside its evidence boundaries.
Save the list snapshot and access date, device and applicant names, submission number, pathway, panel, decision date, decision summary, intended use, AI-enabled function, software version, PCCP status, postmarket records and claim limit.
| Reading object | What it can support | Question to keep visible |
|---|---|---|
| List row | Identifies an AI-enabled marketed device | When was the list accessed? |
| Submission number | Links to the FDA decision record | Which pathway? |
| Labeling | Defines intended use and users | Which software version? |
| Postmarket record | Tracks later safety or change evidence | Which system owns the fact? |
Sources and next reading
Substantive resource claims in this article were checked against the primary pages below. A structured record is a route into evidence; it does not by itself prove a scientific conclusion. Recheck the live record, evidence labels and release context before using it in assessed work or research.
For a classroom enquiry, save a short research receipt containing the question, search terms, selected object, exact identifier, version, access date and the evidence boundary you observed. Add the file or page that supports each statement. This small routine turns browsing into accountable science learning and gives a teacher, classmate or future analyst a route they can independently follow.
Before sharing the conclusion, ask another reader to retrace the route without verbal hints. If they reach a different version, file or interpretation, improve the receipt and narrow the claim. Reproducibility begins with this practical check. It teaches careful reading, supports fair collaboration and keeps uncertainty visible while evidence moves from an online service into notes, analysis and a final explanation.
A useful final review separates five questions. What object was found? Which organisation or repository maintains it? What exact evidence was inspected? Which statement does that evidence support? What remains uncertain? Answer each question in one sentence, then connect every answer to a visible field, file or source page. This review catches common mistakes: citing a collection when a file supplied the value, treating the newest release as the analysed release, confusing permission with scientific suitability, or turning a discovery count into a conclusion. It also gives students a calm way to improve work. A missing detail becomes a next research step instead of an invitation to guess. When the route is complete, the final explanation can stay simple because the identifiers, dates, versions, limitations and supporting evidence are already organised underneath it.
Primary sources
- FDA AI-Enabled Medical Device List — accessed 25 October 2026.
- FDA Artificial Intelligence-Enabled Medical Devices — accessed 25 October 2026.
- FDA Digital Health and AI Glossary — accessed 25 October 2026.
- FDA Medical Device Databases — accessed 25 October 2026.
- FDA Digital Health Center of Excellence — accessed 25 October 2026.