Skip to main content
Everything from Eka MedAI that isn’t a model or a dataset: evaluation tooling, knowledge graphs, document tooling, knowledge bases served over API and MCP, and the reference lists behind our services. For SDKs and agent frameworks across all Eka products, see the Open Source page.

Evaluation

  • KARMA – OpenMedEvalKit – Open-source framework for evaluating medical AI models on Indian clinical tasks; runs every dataset on the Datasets page. GitHub · Launch blog

Knowledge Graphs

  • BODHI – Bharat Ontology for Disease & Healthcare Informatics – Clinician-validated clinical knowledge graphs (conditions ↔ symptoms ↔ specialities; concepts ↔ drugs ↔ lab investigations) in Neo4j, CSV, PyTorch Geometric and RDF. GitHub · Hugging Face · Launch blog

Document Tooling

  • document-pii-redactor – Detects and redacts PII in text and document images; CPU-friendly, trained for Indian names, documents and languages. GitHub · Model weights · Live demo
  • Med Doc Classifier – Lightweight medical document classifier. Model card · Live demo

Knowledge Bases (API & MCP)

Curated Indian medical knowledge, available as web APIs and through the MedAI Tools MCP Server for Claude, ChatGPT, Cursor and other MCP clients. The same knowledge bases are also exposed inside the Eka EMR MCP Server.

Reference Lists

Live lists maintained alongside the services; always check the source rather than a copy.
  • LOINC-linkable lab tests – the 1,400+ investigations our lab report parser links to LOINC. Notion
  • PII categories – the 53 categories detected by the redactor, in the document-pii-redactor README

Where to Find Us