Skillget

MACH

mcp
Guvenlik Denetimi
Uyari
Health Uyari
  • License — License: Apache-2.0
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 5 GitHub stars
Code Gecti
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Gecti
  • Permissions — No dangerous permissions requested

Bu listing icin henuz AI raporu yok.

SUMMARY

Machine-Actionable Catalog for Healthcare

README.md

MACH — Machine-Actionable Catalog for Healthcare

License: Apache-2.0
Data License: CC0-1.0
Validate
DOI
SWH
OpenAIRE

An open, machine-actionable catalog of open-source healthcare software, AI/ML models, clinical standards, datasets, and MCP servers — curated for humans, queryable by agents.

Every entry is a structured JSON-LD record aligned with established open metadata standards (CodeMeta, MLDCAT-AP, DCAT-AP, Croissant), so hospitals, researchers, governments, and AI agents can discover and consume the open healthcare commons.

Project status: Active — CI pipeline live. A project of FORSE-H.

Why MACH?

The open healthcare technology ecosystem is fragmented across hundreds of repositories, blog posts, and HuggingFace model cards — none of which is structured for agent discovery or aligned with open metadata standards.

MACH fills that gap:

  • Structured, evidence-backed entries — every entry includes a curated rationale, clinical domain tags, deployment context, and at least one live evidence URL
  • Three-ring editorial judgement — Adopt / Assess / Caution, data-driven and explained, adapted from the ThoughtWorks Technology Radar
  • Machine-readable exports — MLDCAT-AP 3.0, CodeMeta 3.0, DCAT-AP, Croissant, llms.txt
  • FAIR alignment — stable entry URIs, DOI-archived on Zenodo, ORCID-attributed
  • Agent-native — MCP server and FAIR Data Point RDF/SPARQL endpoint (Phase 2)

Why open source in healthcare?

Open source enables transparency, reuse, and collaborative improvement. These principles matter everywhere, but are especially critical in healthcare, where the cost of opacity is borne by patients and health systems, not vendors.

Who is MACH for?

Audience Use case
Hospitals & health systems Evaluate open-source tools against structured criteria before procurement
Researchers & data scientists Discover FAIR-aligned datasets, models, and pipelines with citable metadata
AI / agent developers Query the catalog programmatically via JSON-LD, MCP server, or SPARQL
Governments & ministries Identify open standards-compliant tooling for national digital health infrastructure

Catalog categories

Category What it covers
Software EHR/EMR, imaging tools, pipeline software, clinical systems
AI / ML Models Clinical LLMs, imaging foundation models, NLP
Datasets Benchmarks, clinical datasets, evaluation suites
MCP Servers FHIR MCP, OMOP MCP, clinical AI interfaces
Data Sources Public health APIs, open data portals, surveillance feeds
Catalogs Other open healthcare catalogs and registries
Specs Interoperability standards and specifications

How to use the catalog

Browse the website:
Visit the GitHub Pages site for a searchable, filterable view of all entries.

Use the data programmatically:
All entries are JSON-LD files in entries/. Clone the repo or fetch individual entries directly:

# Clone
git clone https://github.com/FORSE-H/MACH

# Fetch a single entry
curl https://raw.githubusercontent.com/FORSE-H/MACH/main/entries/software/openmrs.jsonld

Entry structure:
Each entry is a JSON-LD file with fields from CodeMeta, MLDCAT-AP, and the MACH vocabulary:

{
  "@context": "../../data/context/mach.jsonld",
  "identifier": "openmrs",
  "name": "OpenMRS",
  "description": "...",
  "url": "https://openmrs.org",
  "license": "MPL-2.0",
  "mach:judgement": "Adopt",
  "mach:judgementReason": "...",
  "mach:clinicalDomain": ["primary-care", "global-health"],
  "mach:evidence": [...]
}

Vocabulary reference: data/context/mach.jsonld
Taxonomy reference: data/taxonomy/categories.yaml

How it works

GitHub Issue (suggest an entry)
         │
         ▼ Maintainer approves → CI harvests from source systems
         │
         ▼
      DuckDB  ←  Catalog sources (Git · HuggingFace · arXiv · etc.)
         │
         ├── Scoring (Adopt / Assess / Caution)
         ├── Validation
         └── Draft PR → Maintainer reviews → Merges
                  │
                  ▼
         Catalog goes live
         ├── API (JSON-LD · REST · Phase 2)
         ├── GitHub Pages (searchable UI)
         ├── MCP server (Phase 2)
         └── FDP / SPARQL endpoint (Phase 3)

Contributing

Suggest an entry via a GitHub Issue — fill in the template with a name and source URL. The CI pipeline handles harvesting and metadata enrichment; a curator reviews before anything goes live.

Prerequisites for contributors:

  • A GitHub account
  • Familiarity with JSON (entries are JSON-LD files)
  • Access to the project's source URL for the tool you're suggesting

See CONTRIBUTING.md for the full entry checklist, judgement criteria, and conflict-of-interest policy.

License

What Licence
Code (CI pipeline, scripts, site tooling) Apache-2.0
Catalog data (entries/ JSON-LD files) CC0-1.0 — no rights reserved

Cite this catalog

@misc{mach2026,
  title   = {MACH: Machine-Actionable Catalog for Healthcare},
  author  = {Ojha, Priyanka},
  year    = {2026},
  doi     = {10.5281/zenodo.20155320},
  url     = {https://zenodo.org/records/20155320},
  orcid   = {https://orcid.org/0000-0002-6844-6493},
  note    = {CC0-1.0}
}

Contact & community

Acknowledgements & references

Design inspiration

Metadata standards

FAIR scoring

Archival & indexing

License integrity

AI assistance

  • Parts of this project were developed with assistance from Claude (Anthropic). All content has been reviewed and is the intellectual responsibility of the project curator.

Yorumlar (0)

Sonuc bulunamadi