openmed
Local-first Python runtime for clinical named entity recognition and HIPAA-compliant PII de-identification on-device.
Extract structured medical entities and redact patient PII on your local hardware across 21 languages.
OpenMed is an open-source local runtime built for healthcare developers who need to extract clinical entities and de-identify patient records on local devices. It lets you process clinical notes across mobile, desktop, and web platforms without patient data leaving your local network.
What it is
OpenMed provides a Python SDK alongside native client adapters like OpenMedKit, ONNX Runtime Mobile, and Transformers.js bindings. It parses raw medical text to redact sensitive patient PII while running specialized clinical named entity recognition models directly on local hardware.
Who it's for
This tool suits healthtech developers, clinical researchers, and medical IT specialists who build data-processing pipelines requiring strict HIPAA compliance. It is not intended for clinical end-users seeking a non-technical desktop application.
Key features
Local clinical entity extraction
Extract diseases, medications, and clinical signals from unstructured medical text using local Python APIs.
HIPAA PII redaction engine
Redact names, addresses, billing details, and personal identifiers across 33 model-backed PII languages.
Cross-platform hardware bindings
Run models locally on Apple Silicon using MLX, mobile devices via ONNX Runtime, or browsers with Transformers.js.
Extensive medical catalog access
Access over 2,200 model catalog manifest entries mapped for specialized healthcare NLP tasks.
Up and running in 3 steps
Our honest verdict
What's great
- Processes text entirely on-device so patient data never leaves your internal network.
- Supports multi-platform deployment including Python, Apple MLX, iOS, Android, and web browsers.
- Includes pre-configured privacy filters specifically designed for synthetic and real clinical discharge notes.
- Provides multi-lingual support covering 21 clinical languages and 33 PII redaction languages.
Where it falls short
- Requires initial network connectivity to download large model weight artifacts before working offline.
- Requires Python development skills to build and maintain clinical processing workflows.
- Local performance depends heavily on the memory and compute capacity of your targeted edge device.
Pricing & license
OpenMed is completely free to use under the Apache-2.0 open-source license. The repository does not offer a paid hosted tier, so your only expense is the local hardware infrastructure required to run the models.
Open remote AI-training roles
874 open remote AI-training roles, updated November 15, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
Use it, or have us set it up
FAQ
Does OpenMed send patient data to remote cloud servers?
No, inference runs completely on your local device hardware once the required model files are downloaded.
Which platforms can run OpenMed models?
OpenMed supports Python environments, Apple Silicon via MLX, Android via ONNX Runtime Mobile, and web browsers via Transformers.js.
How many languages are supported for PII redaction?
The catalog includes models supporting 33 languages for PII de-identification and 21 languages for clinical tasks.
