Repolode
Submit
maziyarpanahi

maziyarpanahi/openmed

Python SDK for local clinical NER and HIPAA-oriented PII de-identification, running on hardware you control.

View on GitHubWebsite
Stars
5,784
Forks
741
Open issues
480
Last commit
3 hours ago

What is it?

OpenMed is a local-first healthcare AI toolkit. It extracts clinical entities such as diseases, drugs, anatomy and genes from medical text using a catalog of specialized NER models, and it detects and de-identifies personal information such as names, dates, IDs and contact details. The core runtime processes text on your own machine once the required model files are present; model downloads and any remote provider or telemetry paths are optional and separate.

The project provides a Python API, a REST service with Docker support, and batch processing. On Apple Silicon it can run through MLX and a Swift package called OpenMedKit, while Android and browser deployments are supported through ONNX Runtime Mobile and Transformers.js exports. PII extraction covers 35 model-backed languages, and de-identification methods include masking, replacement, hashing and date shifting.

The SDK is released under Apache-2.0. The project states that its tools do not by themselves establish HIPAA compliance, and model and dataset terms must be reviewed before deployment.

Key features

  • 01Clinical NER for diseases, drugs, anatomy and genes using a model catalog
  • 02PII detection and de-identification with mask, replace, hash and date-shift methods
  • 03Runs on CPU or CUDA via Hugging Face, and on Apple Silicon via MLX
  • 04OpenMedKit for Swift/iOS and ONNX exports for Android and browser
  • 05FastAPI REST service and batch processing for many documents
  • 06Supports loading models from a local directory for offline use

Use cases

  • De-identifying clinical notes before sharing them for research or analytics
  • Extracting medication and condition entities from discharge summaries on-premise
  • Running PII redaction inside a hospital network where cloud APIs are not permitted
  • Building an on-device note scanner for iPhone or Android apps

Quick start

  1. 1

    Install the core Python runtime with Hugging Face support

    pip install --upgrade "openmed[hf]"
  2. 2

    Run analyze_text on a clinical sentence in Python

Things to know

  • Model and dataset terms vary and must be reviewed before use.
  • Deployment requires your own validation for privacy behavior and clinical fitness.
  • Some optional paths such as model downloads and remote adapters use the network.

In collections

Run LLMs locallyAndroid apps & phone tools

Topics

#healthcare#clinical-nlp#pii-detection#ner#local-llm#on-device#hipaa#mlx#android#ios

FAQ

What is openmed?
Python SDK for local clinical NER and HIPAA-oriented PII de-identification, running on hardware you control. OpenMed is a local-first healthcare AI toolkit. It extracts clinical entities such as diseases, drugs, anatomy and genes from medical text using a catalog of specialized NER models, and it detects and de-identifies personal information such as names, dates, IDs and contact details. The core runtime processes text on your own machine once the required model files are present; model downloads and any remote provider or telemetry paths are optional and separate.
Is openmed free to use? What license does it have?
openmed is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is openmed still maintained?
Actively maintained. The last commit to openmed was on Nov 7, 2026.
What can I use openmed for?
De-identifying clinical notes before sharing them for research or analytics Extracting medication and condition entities from discharge summaries on-premise Running PII redaction inside a hospital network where cloud APIs are not permitted Building an on-device note scanner for iPhone or Android apps
How do I install openmed?
Install the core Python runtime with Hugging Face support: pip install --upgrade "openmed[hf]" Run analyze_text on a clinical sentence in Python
Where can I get openmed?
The source code and installation instructions are on GitHub: https://github.com/maziyarpanahi/openmed

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects