openmed
OpenMed is a local-priority health AI tool that converts clinical texts directly to structured data in local hardware and maintains confidentiality. The system, which operates without cloud dependency, performs clinical asset recognition (NER) and HIPAA compliant personal identifiable information (PII) anonymization operations. Apple MLX supports more than 2,200 medical models and 21 languages in its infrastructure that can run on ONNX Runtime Mobile and browser environments for Android.
maziyarpanahi · Local AI / Self-Hosted LLM
Common commercial products in this category: ChatGPT, Claude, Gemini
- You can scan clinical notes on local hardware and anonymize personal information.
- With support for MLX processing in Apple Silicon processors, you can extract clinical signals locally.
- On Android devices, you can use Kotlin and ONNX Runtime Mobile to import documents and mask PII.
- ↓ NegativeGitHub Issue
[BUG] Why does OpenMed-NER-PharmaDetect-SuperClinical-434M-mlx classify "PP" and "sugar" as CHEM entities?
“I am evaluating the OpenMed-NER-PharmaDetect-SuperClinical-434M-mlx model for extracting pharmaceutical entities from clinical text. While testing with a sample containing blood sugar values and medication names, I observed that the model identifies expected drug names…”
- • NeutralGitHub Issue
PII 2
“It would be nice if you can do a PII 2 model for Greek language also”