-
Notifications
You must be signed in to change notification settings - Fork 0
Concepts
The domain terms this SDK uses, and why each one exists. If a guide or reference page mentions a term you have not met, it is defined here.
A living model of one patient's body, assembled from their health data and organized by body system.
You never touch raw records. You read and write the twin, which keeps your integration working against one structured model instead of many source formats. In code, a twin is the object returned by dtp.twins.connect(grantToken).
A physiological grouping such as cardiovascular or respiratory.
Systems are how you slice a twin into the part you care about instead of pulling everything at once. twin.systems.get("cardiovascular") returns only that slice, and a grant can restrict which systems you may read.
One timestamped entry on a twin: a lab result, an observation, or a flag you raised.
The clinical fields (the measurement code, its value, its unit, and its system) live inside the event's data object rather than at the top level. That keeps the platform agnostic about which coding system you use, so event.data.code might be a LOINC, SNOMED, or RxNorm code depending on the source.
Consent, encoded. A patient issues a grant that authorizes you to reach exactly one twin, scoped to the body systems and event types they approved.
A grant exists so consent is an explicit, inspectable object rather than something assumed from an API key. Its claims are grantId, twinId, systems, and eventTypes.
The signed proof of a grant: a JWT the patient issues. It is what you pass to dtp.twins.connect().
connect decodes it locally, so the twin.grant claims are available without a network round trip. The platform verifies the token server-side on the first data request. No grant token, no patient data.
Identifies your application to the platform. A key beginning dtp_live_ talks to production; dtp_test_ talks to the sandbox.
On its own a key never unlocks patient data. It says who you are, not what you may see. That separation is why you also need a grant token to reach a twin.
A finding you write back onto a twin as a new event, for example marking an LDL result as above target.
Flagging exists so your integration contributes back rather than only reading. A flag defaults to the eventType "flag", and the grant must permit that event type for the write to succeed.
The clinical-knowledge service behind the platform: drug interactions, concept lookups across SNOMED / RxNorm / LOINC, reference ranges, and phenotype similarity.
The SDK exposes it under dtp.holon so one client covers both a patient's data and the knowledge to interpret it. See its use in Guides and Use-Cases.
Related: Getting-Started, Guides, API-Reference, FAQ.