Health Data Aggregator · India

Architecture diagrams

Nine diagrams generated from Mermaid sources. Click any diagram to open it full size.

Fig 1Fig 2Fig 3Fig 4Fig 5Fig 6aFig 6bFig 7Fig 8

Fig 1 — Ecosystem context

Where the aggregator sits: ABHA gives identity and addressability (not storage); the HIE-CM and Gateway carry consent and routing; HIPs push encrypted FHIR R4 bundles to the aggregator; devices and documents are direct channels; NHCX is participant-only (PAYER / PROVIDER / TPA), so consumer insurance data comes from documents and partnerships.

Fig 1 — Ecosystem context

Fig 2 — Logical architecture

Six layers: channels → edge & policy enforcement (every call passes the consent × purpose policy engine) → acquisition services → normalisation (FHIR ⇄ Health Graph with provenance and trust tier) → intelligence & experience services → India-region data plane with per-person keys.

Fig 2 — Logical architecture

Fig 4 — Document AI pipeline

Intake (upload, camera, WhatsApp, email) → pre-processing → OCR → classification → family-member match → extraction → normalisation (LOINC/SNOMED/ICD/UCUM, brand→molecule) → confidence routing with mandatory confirmation for handwritten medicines.

Fig 4 — Document AI pipeline

Fig 6a — Health Graph: clinical entities

Identity, family and clinical entities. Every node also carries a provenance envelope (source, raw reference, consent IDs, ABDM transaction, confidence, verification, retention).

Fig 6a — Health Graph: clinical entities

Fig 7 — Security zones

Classification-based zones (C4 restricted under per-person KMS keys, C3 clinical, C1 de-identified analytics), SOC with CERT-In 6-hour reporting, no standing workforce access to C4 data.

Fig 7 — Security zones

Fig 8 — Capability roadmap

Legal opinions, DPIA and clinical safety case gate the MVP; four pilot measures gate Phase 2; Phase 3 adds provider SaaS/NHCX, UHI booking and regulated insurance journeys only if licensed.

Fig 8 — Capability roadmap