Roadmap
Four measures across two engines, against every published test case for those measures:
- CMS122, CMS125 and CMS130 on fqm-execution;
- CMS122, CMS125 and CMS165 on HAPI FHIR Clinical Reasoning.
Also working: a Bulk Data loader, a profile-declaration tool, a ConceptMap proposer and a reviewer gate. Both counting modes and the account are working, as shown in the live run report.
The status of every one of the 74 published dQM packages is on the coverage page.
Designed, not yet built
Section titled “Designed, not yet built”Stated so nothing here is read as more than it is:
- the retrieve tracer;
- the fault-injected test decks;
- the bias report.
No rule has yet been clinically signed.
First release
Section titled “First release”- Ten dQMs — CMS2, CMS122, CMS125, CMS130 and CMS165 (the 2026 APP Plus eCQMs), CMS137 (proposed for APP Plus in 2027), and CMS124, CMS131, CMS138 and CMS347. Each has a differential comparison against a Cypress-certified production calculator. See the ten first-release measures.
- The HL7 validator in the pipeline.
- A retrieve tracer that names the empty retrieves and missing elements behind every zero.
- The bias report and an owner-routed feed of findings.
- Signed rule sets for two ambulatory EHR platforms.
- Fault-injected copies of the published test decks for each observed defect class, drawn from 15 EHR platforms.
- A governance charter covering signer scope, liability and how findings about vendors are disclosed.
- Public open-source release of the core, the conformance harness and the test decks.
- Deployment: a public container image — the start of a scalable deployment model.
Then: the full converted set
Section titled “Then: the full converted set”- The full converted set of about 70 measures.
- The first hospital data source, which needs inpatient encounter, medication-administration and discharge data that ambulatory exports do not carry, and a facility partner.
- A scalable deployment model, from one clinic’s server to a national instance, across CMS programs.
- Multi-tenant isolation and a third ambulatory platform.
Who runs it
Section titled “Who runs it”- An organisation with technical staff installs the container.
- A registry or accountable care organisation (ACO) operates it for its members.
- A CMS reference instance would give implementers an arbiter when they disagree, and a channel that returns defects to measure stewards.
Attribution, eligibility and submission stay outside the calculator. They are program rules.
- Engine defects. Both engines have had defects. One — a unit-conversion defect — is now fixed upstream in CQL 5.2.0 and ships in HAPI FHIR 8.12. Fixes not yet accepted upstream are carried as manifest entries. See Prototype results.
- Package collisions. Two packages publish value sets under one canonical URL with different ids. Loading CMS122 beside CMS125 broke 15 of 55 cases. For the first release, the answer is to isolate each measure.
Security and privacy
Section titled “Security and privacy”The calculator runs inside the data owner’s boundary, under HIPAA. Its conformance evidence is synthetic by design and can be shared without patient data. See Security and privacy.
Funding
Section titled “Funding”We show phases, not dates: the pace depends on funding and partners. The first release, the full measure set and any reference instance need funding, to be scoped with CMS.