2026-06-23

Architecture

The Cohort Discovery Service is built on a federated architecture that allows researchers to query pseudonymised health data held by Data Custodians — without that data ever leaving the Custodian's environment.

• Platform Architecture

  ---

  High-level view of how the components fit together.

  Platform Architecture

• Mapping to OMOP

  ---

  How source data is transformed to the OMOP Common Data Model.

  Mapping to OMOP

• Data Governance & Security

  ---

  The controls in place to protect patient data.

  Governance & Security

---

Core principle: data never moves

graph LR
    R[Researcher] -->|"Query (via browser)"| GW[Gateway UI]
    GW --> CDS[Cohort Discovery Service]
    CDS -->|"Task (HTTPS outbound)"| B[Bunny]
    B -->|SQL query| OMOP[(OMOP DB)]
    OMOP --> B
    B -->|"Aggregate count only"| CDS
    CDS --> GW
    GW --> R

    subgraph Custodian Infrastructure
        B
        OMOP
    end

The query tool (e.g. Bunny) makes only outbound requests — no inbound connections are required through the Custodian's firewall. Only aggregate, disclosure-controlled counts are returned.