One patient · one record

Four sites. One attested record.

The brands look separate — a triage site, a class, a coach, a surgeon tool. They are four roles in a single patient record. Follow one episode from first symptom to final code: at every step the model proposes, a check that is not the model verifies, a named human signs, and a hash seals it. The receipts chain into one longitudinal, auditable record.

Synthetic patient · demonstration only · not coding or medical advice
PatientR. Vance, 64 F
ConditionR knee OA · K-L grade 4
PlanMedicare Advantage
EpisodeMSK-2026-0148
01Find + triagearthritisrisk.com

Mechanical, not inflammatory.

A two-minute screener sorts the knee pain: mechanical OA, not an inflammatory arthritis — the right patient for a conservative-first pathway, not a straight-to-surgery referral. Routed accordingly.

◆ RECEIPT   triage · checked by ICD-10 rule set · M17.11 R primary OA of knee · attested R. Vance (patient) · sha256 a1f3…9c02
↓
02Learnjointclass.com

She understands the options before anyone operates.

The knee-OA class — what the diagnosis means, what conservative care can and can’t do, when surgery is warranted — completed before a surgical decision. Informed, on the record.

◆ RECEIPT   education · module “Knee OA & your options” completed · attested R. Vance (patient) · sha256 7b40…11de
↓
03Track + measurejointcoach.com

Twelve weeks of conservative care — measured, not assumed.

Home-exercise RTM with a validated outcome captured at baseline and follow-up: KOOS-JR 42 → 44 after 12 weeks — minimal functional gain despite documented adherence. Conservative care given a fair, measured trial; now a surgical candidate on evidence, not a hunch. This is the PRO the value-based contract pays on.

◆ RECEIPT   PROM · checked by KOOS-JR scoring · baseline 42 → 44 (Δ+2) · adherence 81% · attested PT, license on file · sha256 c9a2…4f70
↓
04Operate · code · attestsurgeonvalue.com

The surgeon codes it — and an engine, not the model, checks it.

TKA performed. The op-note reads as 27447; the model proposes the codes and quotes the supporting line, then the deterministic CMS NCCI engine checks each one — the model does not grade its own work — and the surgeon signs. One code came back a caution to resolve before submission.

◆ RECEIPT   coding · checked by CMS NCCI (Q1 2025) · 27447 clean · G2211 not-pair-checkable · attested surgeon, license on file · sha256 e75b…81c0
The record · solvinghealth · comfortcard

Five steps. Four receipts. One episode record.

Nothing above lives in a silo. Each step emitted a receipt — what was proposed, what external source verified it, who signed, and a hash that seals it — and they chain into a single longitudinal record of the whole episode, from first symptom to final code.

triage a1f3 education 7b40 PROM c9a2 coding e75b → episode MSK-2026-0148

The model read; an external check said yes or no at every step; a named human signed; a hash made it tamper-evident. PHI never left the record — the proof travels, the note stays home.

The short codes in the chain above are illustrative. This box makes a real SHA-256 of the coding receipt on your device. Change one character and the whole fingerprint changes: that is what tamper-evident means. Nothing is sent.

Needs JavaScript; with it off, any SHA-256 tool on your own computer gives the same answer for the same text.

Who buys this

The party at risk for the whole episode.

Not the surgeon (buys coding) or the PT (buys tracking). The risk-bearing entity — the ACO, the ACCESS participant, the bundle-holder, the provider-owned plan — is the only one who needs the entire record:

Where each step runs

01 triage on arthritisrisk.com · 02 the class on jointclass.com · 03 tracking on jointcoach.com · 04 coding on surgeonvalue.com.

This is the harness, walked end to end.

The homepage shows the network. This shows one patient moving through it — every output seen, every signer controlled, the whole audit trail owned. Where does that check live in your stack?