Skip to main content
CO-LUMINATECausal evidence platformProject WorkspaceChecking access

How it works

An iterative path from a causal question to governed use.

CO-LUMINATE separates the scientific workflow from project delivery. Work can return to an earlier decision whenever data readiness, measurement, analysis, interpretation, or release review identifies a limitation.

Workflow
Approved foundation definition
Structure
Iterative, not a project timeline
Implementation status
Depends on the bounded implementation

Scientific workflow

Each transition creates a reviewable output.

Completing a step records workflow state only. It does not prove that the output is correct or that the next approval gate has been met.

  1. 01

    Co-creation and intended use

    Define the causal concern, decision context, relevant experience, interpretation risks, and bounded use.

  2. 02

    Data readiness

    Assess provenance, population scope, linkage, observation, missingness, measurement, privacy, and uncertainty.

  3. 03

    Causal and measurement specification

    Develop the causal story, pathway inventory, diagram, endpoint, time ordering, and measurement plan.

  4. 04

    Analysis specification and execution

    Define the estimand, data requirements, methods, diagnostics, sensitivity work, and reproducible run.

  5. 05

    Interpretation and challenge

    Review findings against assumptions, uncertainty, measurement limits, lived-experience knowledge, and alternatives.

  6. 06

    Translation and governed release

    Prepare bounded learning, research, implementation, or policy outputs and apply the required review gates.

  7. 07

    Feedback, learning, and adaptation

    Record what was learned, revise the intended use or implementation, and return to the relevant earlier step.

Layer 0

A cross-cutting data-readiness gate.

Layer 0 asks whether provenance, linkage, observation history, missingness, transformations, measurement, privacy, and uncertainty are adequately understood for the proposed use. It can be revisited whenever the question, population, outcome, data, time window, or method changes.

Concrete variables, rules, thresholds, and configurations belong to the applicable scientific implementation. They are not universal platform defaults.

Capability architecture

Five capability groups, not five project phases.

Layer 1

Life-course reconstruction

Reconstruct and audit histories, timing, linkage, and observation gaps.

Layer 2

Outcome harmonisation

Define endpoints, scoring provenance, timing, validation, and portability limits.

Layer 3

Causal story and pathway

Translate co-created knowledge into reviewable pathway and measurement questions.

Layer 4

Analysis engine

Specify estimands, analysis requirements, diagnostics, sensitivity work, and auditable outputs.

Layer 5

Website, tools, and reporting

Govern learning, access, reporting, release, and public or controlled presentation.

Decision authority

Review gates follow the object, version, scope, and intended use.

See who reviews each decision domain