openmed: what it is, what problem it solves & why it's gaining traction

What it solves

OpenMed provides a local-first approach to healthcare AI, allowing users to process clinical text without sending sensitive patient data to cloud vendors. It solves the problem of privacy and HIPAA compliance by enabling entity extraction and PII (Personally Identifiable Information) de-identification to run entirely on the user's own hardware.

How it works

OpenMed uses a curated registry of over 1,000 specialized biomedical and clinical NER (Named Entity Recognition) models. It supports multiple backends, including PyTorch (for CPU and CUDA) and Apple's MLX for acceleration on Apple Silicon. For iOS and macOS apps, it provides a native Swift library called OpenMedKit. The system can extract medical entities (like diseases and drugs) or detect and redact PII using various methods such as masking, hashing, or replacing with fake data.

Who it’s for

It is designed for healthcare providers, medical researchers, and app developers who need to process clinical text while maintaining strict data privacy and avoiding vendor lock-in.

Highlights

• 100% On-Device: Clinical text never leaves the device or local network.
• Extensive Model Library: Access to 1,000+ specialized medical models across 15 languages.
• PII De-identification: HIPAA-aware redaction of 18 Safe Harbor identifiers with smart entity merging.
• Cross-Platform Support: Runs on Linux, Windows, macOS, and natively on iOS/iPadOS via OpenMedKit.
• MLX Acceleration: Significant speedups (24-33x) on Apple Silicon compared to CPU PyTorch.
• Flexible Deployment: Available as a Python API, a Dockerized REST service, or batch processing pipelines.