Pre-Visit Check-in: Closing the Loop
CMS Health Tech EcosystemKill the Clipboard · August 2026

Pre-Visit Check-in: Closing the Loop

How a provider asks, how a patient answers, and how the visit moves on.

advance with or space
02Where this came from

An update from the breakout

We ran three breakout calls on pre-visit check-in design, covering workflow, timing, and candidate protocols. Two proposals are on the table. Today is an update, not a decision ask:

Lay out the problem

Describe it in terms everyone on this call can evaluate, without breakout shorthand.

Walk through the two models

What the proposals share, where they differ, and what each one asks builders to build.

Preview the fall demos

Working, multi-vendor demonstrations are within reach this fall, without waiting on anyone who isn't ready.

03The problem

Practices say what data they expect, and when. Apps help patients select and share.

Maria books a Thursday appointment. Her practice needs different things from her at different moments, not one blob of data on arrival:

At booking
  • Insurance card
  • Medication list
  • History, allergies
  • Reason for visit
Day before
  • PHQ-2 depression screen
  • Anything changed since booking?

Answers stay fresh, and a positive screen doesn't sit unread for a week.

At arrival
  • Copay
  • Consents
  • "You're checked in."
04Common ground

Points of consensus

05The design space

The models differ on where the response goes

A portal prompt, a texted link, a kiosk QR code, or an app that notices the appointment: any of these can start the flow, under either model. The choice worth making carefully is what happens when the patient answers.

Entry: shared by both models

Each of these lands the patient in the same check-in flow.

portal messageemail / SMS linkQR at the deskapp discovers appointment

Response: where they split

The Autofill model returns the response to the page that asked, and the visit continues there. The API Submission model has the app write to a server, and the exchange ends in the app.

portal message email / SMS link QR at the desk app discovery one check-in flow what happens next: where the models differ
Entry points all funnel into the same flow; the models differ on what happens after the patient answers.
06Walkthrough

The Autofill model

The provider's check-in page issues the request through the browser, the same credential surface phones already use to present a driver's license. The patient approves item by item in an app they choose, and the answer returns to the page, which continues to copay and consents. A kiosk QR or texted link lands walk-ins in the same exchange, so pre-visit and on-site check-in share one protocol.

provider check-in page patient's chosen app request: items + shapes signed artifacts back → copay · consents · "you're checked in"
Request out, response back, workflow continues: no accounts created, no app pre-registered.

Backend integration: out of scope

Deliberately: the receiving page forwards results into whatever the practice already has. FHIR-ready systems POST FHIR; others use their existing intake pipeline. Start now, build toward FHIR everywhere instead of blocking on it.

The mdoc transport

Rides the W3C Digital Credentials API over ISO mdoc: rails that ship on iOS and Android today, maintained by platform vendors for licenses and IDs (and drawing EHR interest beyond check-in). The machinery is heavier than plain JSON; a shared browser SDK handles it. Size limits and consent UX still need real testing.

07Walkthrough

The API Submission model

The patient's app writes the data to a server: the EHR's FHIR endpoint, or one of the CMS-aligned networks in between, using a registered client identity and an explicit authorization step. One backend serving many use cases is the appeal; the exchange ends in the app.

patient's app CMS-aligned networks (~40) EHR FHIR API (write) join + credential, per app, per network write path: not defined today patient parked here: copay, consents unfinished provider check-in page never hears back
Networks in the middle split pairwise setup into two halves, app-to-network and network-to-EHR. Both still have to exist; the write path is undefined.

What has to be specified before this works

  • Which endpoint receives the write: the EHR, or a network? Which one?
  • How does each app get a client registration there, at scale?
  • What does the patient's authorization step look like?
  • How does the provider learn the data arrived and match it to the visit?
  • How does the patient get back to copay and consents?
08Side by side

Scoring the two models

Requirements from the common ground, plus the practical questions raised in the walkthroughs.

RequirementAutofillAPI Submission
FINISHPatient finishes check-in in the provider's flow Yes: the response returns to the page mid-flow Needs added return machinery to be designed
PATIENT-CHOICEPatients share only the items they choose Yes: per-item consent in the credential picker Yes: per-item choice in the app
WORKS-WITHOUT-APPWorks for patients with no app Yes: the portal form works on its own, and can optionally offer an app link-out Yes: portal fallback
ANY-APPAny app the patient chooses can answer Yes: no enrollment, no credentials issued to the app Each app registers and is vetted with each network or write endpoint
ON-SITESame protocol at the front desk Yes: a kiosk QR or texted link lands in the same flow Would need its own walk-in design
BACKENDWhat EHR and provider systems must build first Nothing standardized: each EHR manages its own backend design (FHIR included) Write API, client registration, and authorization at every system, or a new network write path
DEMOEnd-to-end demo by Thanksgiving Yes, with volunteer wallets and one or more EHRs Writes to a test endpoint, without integration into a provider check-in flow
TRUSTWhere trust lives In the existing patient session + optionally signed artifacts + patient attestation In API and identity credentials held by each app
09What's next

Demos by Thanksgiving

This fall · volunteers

A reference check-in surface backed by one or more EHRs, answered by wallets from the patient apps and platform vendors ready to move now. The reference surface forwards results into its backend, showing the backend-integration pattern live.

Nothing blocks on EHR or provider systems; they can join at whatever depth suits their timelines, or just watch.

As EHR and provider systems come aboard

First-party portal integration, testing against real sandboxes, and a Connectathon with the full membership.

Josh's take

We should put our energy into the Autofill wire protocol, with app-discovered entry and per-system backend integration documented as patterns on top. I'm aiming to have running demos to react to by Thanksgiving.

10Discussion

How can I help?

A1Appendix · the trust question

Trust attaches to the data, not the app

"Shouldn't the app be vetted?" is the natural question about the Autofill model, where any app can answer. The answer is that the app is a carrier, and the things worth trusting travel with the data:

Signed artifacts stay signed

A SMART Health Card carries its issuer's signature no matter which app presents it. The receiving system verifies the signature, not the messenger.

Patient answers are patient answers

Unsigned responses are the patient's own statements, exactly as trustworthy as their handwriting on the clipboard was. Nobody ever vetted the pen.

Nothing to revoke

The app never receives backend credentials, so there is no standing access to vet, monitor, or revoke. App listings and identity assurance solve account access, which is a different problem.

The API Submission model needs app vetting because it hands apps API credentials. That's a requirement the model creates, not one the use case has.

A2Appendix · where the work stands

The ingestion half is already built

EHRs that support the Kill the Clipboard QR-code flow already verify and ingest SMART Health Cards. The Autofill response carries those same artifacts, so it lands in the pipeline those systems already run. This is a new front door on existing plumbing, not a new stack.

What the QR work taught us

Front-desk scanning never took hold, because site-by-site hardware and workflow change doesn't happen. The verification and ingestion backends shipped anyway, and they carry over unchanged.

Draft spec: SMART Health Check-in 1.0

A normative, transport-neutral clinical request/response model, plus a same-device binding over the W3C Digital Credentials API (ISO mdoc). FHIR-native selectors, inline questionnaires, per-item status, SMART Health Card and raw FHIR artifacts.

Running code

Android wallet, browser verifier SDK, live web demo, and a kiosk handoff demo, all validated against real captured byte fixtures by test suites in three languages.