Skip to content
@EvidenceOS

EvidenceOS, Inc

Clinical evidence automation for health AI — evaluation, safety, and governance infrastructure

EvidenceOS

Building the operating system for sovereign health intelligence.

We automate the production of trustworthy clinical evidence for health AI systems — evaluation, safety, and governance infrastructure with mathematical assurances. Starting with neurodiagnostic AI (TBI first vertical), expanding to every domain where clinical AI needs to prove it works before it deploys.

"82% of clinicians know about TBI blood biomarkers. 6% use them. The gap isn't science — it's evidence infrastructure."


Methodology — What We Contribute

Open-source research packages powering clinical validation intelligence. Built on evidence from $165M+ of publicly funded research (TRACK-TBI, CENTER-TBI, InTBIR).

Package What It Does Language
evidenceos-multiverse 12-axis specification curve analysis — systematically tests model robustness across thousands of defensible analytical choices Python
evidenceos-conformal Distribution-free prediction intervals (MondrianCQR) — models know when they don't know Python
evidenceos-knowledge Knowledge graph reasoning (GraphRAG + DGCI) — explainable clinical evidence synthesis Python
evidenceos-tabular Clinical prediction model pipeline — loading, harmonization, and feature engineering for TBI datasets Python
evidenceos-ontology NINDS TBI Common Data Elements — 13 machine-readable YAML schemas (7,780 lines) YAML
evidenceos-acquisition Evidence acquisition pipeline — automated search, screening, and knowledge graph construction Python
evidenceos-trajectories Neural ODE biomarker trajectory modeling — temporal disease progression Python
evidenceos-living-review Living systematic review automation — continuous evidence synthesis Python
evidenceos-bench Clinical AI benchmark suite — TRIPOD+AI compliance and model comparison Python

All 10 packages: evidenceos-research (Apache-2.0)

Methodological foundations: Specification curve analysis (Simonsohn et al. 2020), conformal prediction (Vovk et al. 2005), TRIPOD+AI (Collins et al. 2024), Riley sample size criteria (Riley et al. 2020).


Ecosystem — How It Fits Together

Four integrated product families, one evidence foundation. Each product draws from the same Canonical Evidence Library (CEL) — versioned, DOI-addressable analytical artifacts that serve as the evidentiary ground truth.

                         ┌─────────────────────────┐
                         │   Evidence Foundation    │
                         │  (CEL + Knowledge Graph  │
                         │   + Ontology + LSRs)     │
                         └────────┬────────────────┘
                                  │
            ┌─────────────────────┼─────────────────────┐
            │                     │                     │
   ┌────────▼────────┐  ┌────────▼────────┐  ┌────────▼────────┐
   │ Clinic-in-a-Box │  │  Lab-in-a-Box   │  │University-in-Box│
   │                 │  │                 │  │                 │
   │ BRIDGE-TBI      │  │ Research-as-a-  │  │ RAIGH Academy   │
   │ Aya Clinician   │  │ Service         │  │ AI workforce    │
   │ TeleCare Africa │  │ Evidence        │  │ development for │
   │                 │  │ Capsules        │  │ global health   │
   │ For: Clinicians │  │ For: Researchers│  │ For: Universities│
   └─────────────────┘  └─────────────────┘  └─────────────────┘
                                  │
                         ┌────────▼────────┐
                         │AIDA Infrastructure│
                         │                 │
                         │ Sovereign data   │
                         │ for governments  │
                         │                 │
                         │ For: Ministries  │
                         │ of Health        │
                         └─────────────────┘

Current focus: TBI (first vertical). The methodology is domain-agnostic — each new clinical domain requires a domain ontology + constraints + domain experts, not new infrastructure. Expansion roadmap: Sepsis → Cardiovascular → Stroke → Surgical Risk.


Open Resources

Community assets for the health AI ecosystem:

Resource Description
awesome-health-ai-skills 17 executable health AI skills organized in 7 bundles — the Elements of Health AI
awesome-health-ai-regulations Global regulatory frameworks (FDA, EU AI Act, Africa, Asia-Pacific) with structured JSON data
awesome-clinical-prediction-models Living registry of clinical prediction models with YAML-structured evaluations
awesome-ai-evaluation-metrics Cross-modality evaluation taxonomy for health AI systems

SDG Alignment: SDG 3 (Good Health — Targets 3.4, 3.8, 3.d) · SDG 4 (Quality Education — Target 4.4) · SDG 9 (Innovation — Target 9.5) · SDG 17 (Partnerships — Target 17.6)


Governance & Trust

EvidenceOS operates through four legal entities with distinct roles — separating commercial product development from independent evidence production and open-source stewardship:

Entity Role Jurisdiction
EvidenceOS Inc. Product company, IP holder US (Delaware)
EvidenceOS OÜ EU operations, CE Mark pathway Estonia
CEAIH Independent evidence production institute, grant applicant (NIH, Gates, Wellcome) US 501(c)(3)
PAATHI Open-source stewardship, LMIC access, continental AI governance Rwanda (CLG)

Data policy: This organization contains NO real patient data. All included data is synthetically generated. Real patient data is handled in separate, sovereign environments compliant with HIPAA, GDPR, and national data laws. See DATA_POLICY.md for details.

Licensing: Research packages are Apache-2.0. Awesome-lists are CC-BY-4.0. Platform code follows dual licensing — see individual repo LICENSE files.


Get Involved

Researchers & Academics

Clinicians & Health Systems

Governments & Development Partners

Builders & Contributors


Website · Research Packages · Platform · Contact

Apache-2.0 · Built in Delaware, Turku, Kigali, and everywhere clinical AI needs to prove it works.

Popular repositories Loading

  1. ECT-React-samples ECT-React-samples Public

    Forked from ICD-API/ECT-React-samples

    Embedded Coding Tool (ECT) samples in the React framework

    TypeScript

  2. Python-samples Python-samples Public

    Forked from ICD-API/Python-samples

    ICD-API samples in Python

    Python

  3. ECT-JavaScript-samples ECT-JavaScript-samples Public

    Forked from ICD-API/ECT-JavaScript-samples

    Embedded Coding Tool (ECT) samples in Vanilla JavaScript

    HTML

  4. awesome-african-digital-health awesome-african-digital-health Public

    A curated list of digital health standards, frameworks, open-source platforms, and policy resources for building sovereign health AI infrastructure across Africa. Maintained by EvidenceOS & PAATHI.

  5. awesome-dpi-infrastructure awesome-dpi-infrastructure Public

    Awesome list of Digital Public Infrastructure (DPI) platforms, protocols, identity systems, and governance frameworks for health systems

  6. awesome-health-ai-regulations awesome-health-ai-regulations Public

    Awesome list of health AI regulatory frameworks, approved devices, and compliance resources with structured JSON data

Repositories

Showing 10 of 10 repositories

People

This organization has no public members. You must be a member to see who’s a part of this organization.

Top languages

Loading…

Most used topics

Loading…