Problem: 14 specialties had incompatible intake processes — orthopaedics needed body diagram annotations, psychiatry needed PHQ-9 scoring, paediatrics needed guardian consent flows. No unified system existed. Intake data lived in PDFs, fax, and disconnected EMR silos.
Solution: Discovery sprint mapped intake quirks per specialty. A shared form engine with specialty-specific field sets driven by JSON configuration kept the data model and audit trail consistent. AWS HealthLake stored FHIR R4 resources — every form submission translated to a structured FHIR bundle via the HealthLake API. Clinician dashboard queried HealthLake FHIR resources directly — no PHI touched the application database. Patient mobile app used AWS Amplify with Cognito — no PHI stored on device.
Technology: Next.js · React Native · AWS HealthLake · FHIR R4 · Cognito · Amplify · Terraform
Optimisation pattern: specialty-specific-pdfs-to-json-configured-shared-form-engine-on-fhir
Outcomes:
First clinic live in 6 weeks, all 14 specialties live in 14 weeks. Zero PHI incidents through SOC 2 Type II audit. 63% reduction in intake-to-appointment time. Clinician dashboard satisfaction score: 4.7/5 in post-launch survey.