// the find
beda-software/fhir-emr
EMR based on FHIR
A FHIR R4-based EMR frontend in TypeScript that uses Questionnaire/QuestionnaireResponse resources (SDC IG) as the primary UI primitive for forms, clinical notes, and CRUD operations. Aimed at healthcare organizations and developers building clinical applications who want a head start on FHIR-native workflows rather than bolting FHIR onto an existing app.
- Genuinely FHIR-native architecture: forms, extractions, and data models are all FHIR resources (Questionnaires, Mappings, etc.), not just a wrapper that calls a FHIR API for storage.
- Extensive seed data with real clinical questionnaires (PHQ-9, GAD-7, vitals, immunizations, prescriptions, consent forms) means you can actually see the system doing clinical work immediately rather than staring at empty screens.
- Good separation of concerns with an emr-template repo and config package, making it feasible to white-label or extend without forking core code.
- CI, Storybook, Chromatic, and some test coverage are present — rare for healthcare open-source projects which often have zero automated testing.
- Primary backend is Aidbox, a commercial product requiring a paid license for any PHI-handling production deployment. Several features use Aidbox-specific APIs (AidboxQuery, Mapping resources) that don't exist in open FHIR servers, so swapping to HAPI or Azure FHIR is non-trivial despite the README's optimistic 'just needs FHIR API' claim.
- Only 120 stars and the project status is still listed as 'development' after multiple years — community is thin, meaning bug reports may go unanswered and you're largely dependent on Beda Software's roadmap priorities.
- Security model is entirely Aidbox AccessPolicy YAML files, with no documented threat model, audit logging strategy, or guidance on HIPAA/GDPR compliance obligations — dangerous for anyone deploying this thinking 'open source EMR = compliant EMR'.
- Test coverage is shallow: a handful of files in __tests__ against a large feature surface covering scheduling, billing, telemedicine, and medication management. Any non-trivial customization risks silent regressions.