Runtime APIs / Interfaces

DT4H exposes conceptual interfaces across modeling, state, and execution layers.

The interface model separates signal ingestion, cohort resolution, Twin state, StateK computation, calibration, and SETPOINT execution.

Interface diagram #

/signalsInput Layer

Normalize biomarker, behavior, clinical, and self-report signals

/twinModel Layer

Cohorts · reference priors · Twin state · calibration

/statekComputation Layer

State · confidence · transition readiness

/setpointExecution Layer

Protocols · practices · outcomes · feedback

Conceptual interface groups #

01

/signals

Normalize biomarkers, behavior, clinical context, and self-report.

02

/cohorts

Resolve probabilistic cohort context and reference-human priors.

03

/twin

Initialize, retrieve, and update the longitudinal individual model.

04

/statek

Compute current state, confidence, and transition readiness.

05

/calibration

Apply evidence, outcomes, and recalibration updates.

06

/setpoint

Expose execution-ready state to protocols, practices, and feedback loops.

Interface boundary rules #

/signalsInput normalization

Input systems should not make model or clinical claims.

/cohortsContext resolution

Cohort interfaces expose probabilistic context, not fixed identity.

/twinModel state

The Twin is an individualized longitudinal model, not the execution layer.

/statekState computation

StateK computes readiness and transition logic for downstream execution.

Runtime event flow #

Ingest signalSignals enter through normalized runtime event paths.
Resolve cohortCohort intelligence updates context and reference priors.
Update TwinThe individualized model incorporates evidence and calibration state.
Compute StateKState, confidence, trajectory, and transition logic are derived.
Expose to SETPOINTExecution systems receive bounded state for protocols and practices.
Return outcomeResults become evidence for recalibration.

Implementation notes #

Keep API boundaries conceptual until implementation hardens

The current routes describe architecture interfaces, not final public API contracts.

Version model-facing interfaces early

Signal, cohort, Twin, StateK, and calibration contracts should evolve independently.

Never expose raw clinical authority through runtime APIs

Interfaces should return bounded modeling and state outputs, not diagnosis or treatment decisions.

LayerRuntime Interfaces
StatusActive Draft
SystemDT4H / StateK / SETPOINT
BoundaryInfrastructure, not diagnosis
System lineageDT4HTwinStateKSETPOINTOutcomesRecalibration
Infrastructure boundaryDT4H models cohorts, Twins, calibration, and runtime state. It does not diagnose, prescribe, or replace licensed clinical judgment.
Document statusInfrastructure draft
Last updatedMay 2026
Applies toDT4H.ai / AvatarK.ai ecosystem