Turn clinical text into structured insight with one line of code.
Entity extraction, PII de-identification, and 1,000+ specialized medical models that run entirely on
your own hardware — from a one-liner in Python to a native Swift app on iPhone, powered by Apple MLX.
No cloud. No vendor lock-in. No patient data leaving your network.
1,000+ models · 12 languages · 247 PII checkpoints · 100% on-device · Apache-2.0
English · 简体中文 · Español · Français · Deutsch · Italiano · Português · Nederlands · العربية · हिन्दी · తెలుగు · 日本語 · Türkçe · فارسی
OpenMed runs entirely on the device — clinical text never leaves it. Here it is on iPhone, fully offline:
On iPhone via OpenMedKit — scan a clinical note, de-identify it, and extract clinical signals, all locally with Apple MLX. Nothing is uploaded.
Real-time PII de-identification — the Nemotron Privacy Filter redacting names, addresses, IDs, and billing data from a clinical discharge packet, entirely on-device. (All values shown are synthetic.)
from openmed import analyze_text
result = analyze_text(
"Patient started on imatinib for chronic myeloid leukemia.",
model_name="disease_detection_superclinical",
)
for entity in result.entities:
print(f"{entity.label:<12} {entity.text:<28} {entity.confidence:.2f}")
# DISEASE chronic myeloid leukemia 0.98
# DRUG imatinib 0.95A state-of-the-art clinical NER model running locally — no API key, no network call.
| OpenMed | Cloud medical APIs | |
|---|---|---|
| Runs on your device / servers | ✅ | ❌ |
| Patient data leaves your network | Never | Sent to the vendor |
| Cost | Free & open-source | Per-call pricing |
| Specialized medical models | 1,000+ | Limited |
| Languages | 12+ | Varies |
| Offline / air-gapped | ✅ | ❌ |
| Apple Silicon (MLX) acceleration | ✅ | n/a |
| Native iOS / macOS apps | ✅ via OpenMedKit | ❌ |
| Vendor lock-in | None — Apache-2.0 | Yes |
- Specialized models — 1,000+ curated biomedical & clinical models, many outperforming proprietary stacks.
- HIPAA-aware de-identification — all 18 Safe Harbor identifiers, smart entity merging, format-preserving fakes.
- Runs everywhere — CPU, CUDA, Apple Silicon (MLX), and natively in iOS/macOS apps via OpenMedKit.
- One-line deployment — Python API, Dockerized REST service, or batch pipelines.
- Zero lock-in — Apache-2.0, your infrastructure, your data.
OpenMed is built to run where your data already lives. On Apple hardware it accelerates with MLX, and it ships straight into iPhone, iPad, and Mac apps through OpenMedKit — so PII detection and clinical extraction happen fully offline, on the device.
// Add OpenMedKit to your app
dependencies: [
.package(url: "https://github.com/maziyarpanahi/openmed.git", from: "1.5.5"),
]- MLX runtime for PII token classification, the Privacy Filter family, and experimental GLiNER-family zero-shot tasks — with a CoreML fallback path.
- One model name, every platform — MLX model names automatically fall back to the matching PyTorch checkpoint on non-Apple hardware.
- Python on Apple Silicon too:
pip install "openmed[mlx]".
Guides: MLX backend · OpenMedKit (Swift) · CoreML export
MLX on Apple Silicon: 24–33× faster than CPU PyTorch for the Privacy Filter — median latency per inference step, lower is better.
flowchart LR
A["Clinical text"] --> B["OpenMed<br/>(100% on-device)"]
B --> C["Medical entities"]
B --> D["PII detected"]
B --> E["De-identified text"]
style B fill:#0D6E6E,stroke:#0A5656,stroke-width:2px,color:#ffffff
style C fill:#D6EBEB,stroke:#0D6E6E,color:#0E1116
style D fill:#F7DCD8,stroke:#C5453A,color:#0E1116
style E fill:#F5E27A,stroke:#A9A088,color:#0E1116
# Core + Hugging Face runtime (Linux, macOS, Windows; CPU or CUDA)
pip install "openmed[hf]"
# Add the REST service
pip install "openmed[hf,service]"
# Apple Silicon acceleration (MLX)
pip install "openmed[mlx]"|
Python API from openmed import analyze_text
analyze_text(
"Patient received 75mg "
"clopidogrel for NSTEMI.",
model_name=
"pharma_detection_superclinical",
) |
REST service uvicorn openmed.service.app:app \
--host 0.0.0.0 --port 8080
|
Batch from openmed import BatchProcessor
p = BatchProcessor(
model_name=
"disease_detection_superclinical",
group_entities=True,
)
p.process_texts([...]) |
Offline / air-gapped? Point model_name (or model_id) at a local directory and OpenMed loads it without contacting the Hugging Face Hub:
from openmed import OpenMedConfig, analyze_text
result = analyze_text(
"Patient presents with chronic myeloid leukemia and Type 2 diabetes.",
model_id="./models/OpenMed-NER-DiseaseDetect-SuperClinical-434M",
config=OpenMedConfig(device="cpu"),
)A curated registry of specialized medical NER models — browse the full catalog.
| Model | Specialization | Entity types | Size |
|---|---|---|---|
disease_detection_superclinical |
Disease & conditions | DISEASE, CONDITION, DIAGNOSIS | 434M |
pharma_detection_superclinical |
Drugs & medications | DRUG, MEDICATION, TREATMENT | 434M |
pii_superclinical_large |
PII & de-identification | NAME, DATE, SSN, PHONE, EMAIL, ADDRESS | 434M |
anatomy_detection_electramed |
Anatomy & body parts | ANATOMY, ORGAN, BODY_PART | 109M |
gene_detection_genecorpus |
Genes & proteins | GENE, PROTEIN | 109M |
from openmed import extract_pii, deidentify
text = "Patient: John Doe, DOB: 01/15/1970, SSN: 123-45-6789"
# Extract PII with smart merging (prevents tokenization fragmentation)
result = extract_pii(text, model_name="pii_superclinical_large", use_smart_merging=True)
# De-identify with the method you need
deidentify(text, method="mask") # [NAME], [DATE]
deidentify(text, method="replace") # Faker-backed, locale-aware, format-preserving fakes
deidentify(text, method="hash") # Cryptographic hashing
deidentify(text, method="shift_dates", date_shift_days=180)- Smart entity merging keeps
01/15/1970whole instead of fragmenting it. - Faker-backed obfuscation with custom clinical-ID providers (CPF, CNPJ, BSN, NIR, Codice Fiscale, NIE, Aadhaar, Steuer-ID, NPI).
- HIPAA: all 18 Safe Harbor identifiers, configurable confidence thresholds.
- Batch PII (v1.5.5): extract or de-identify across many documents with
BatchProcessor(operation="extract_pii" | "deidentify", batch_size=16).
Batch processing — up to 3.3× higher throughput on CPU and 2.2× on MLX vs. one document at a time.
Complete PII notebook · Smart merging · Anonymization
Privacy Filter family — three model families on the OpenAI Privacy Filter architecture
Same model code (gpt-oss-style sparse-MoE transformer with local attention, sink tokens, RoPE+YaRN, tiktoken o200k_base), different training data. All route through the same extract_pii() / deidentify() API — only model_name= changes.
| Variant | PyTorch (CPU + CUDA) | MLX (Apple Silicon) | MLX 8-bit |
|---|---|---|---|
| OpenAI Privacy Filter | openai/privacy-filter |
OpenMed/privacy-filter-mlx |
…-mlx-8bit |
| Nemotron-PII fine-tune | OpenMed/privacy-filter-nemotron |
…-nemotron-mlx |
…-nemotron-mlx-8bit |
| OpenMed Multilingual | OpenMed/privacy-filter-multilingual |
…-multilingual-mlx |
…-multilingual-mlx-8bit |
from openmed import extract_pii
text = "Patient Sarah Connor (DOB: 03/15/1985) at MRN 4471882."
extract_pii(text, model_name="openai/privacy-filter") # PyTorch baseline
extract_pii(text, model_name="OpenMed/privacy-filter-nemotron") # same code, different weights
extract_pii(text, model_name="OpenMed/privacy-filter-mlx") # Apple Silicon (MLX)On non-Apple-Silicon hosts, MLX model names are automatically substituted with the matching PyTorch checkpoint (with a one-time warning) — ship one model name, run anywhere. See Privacy Filter architecture & backend routing.
Extraction and de-identification across en, fr, de, it, es, nl, hi, te, pt, ar, ja, and tr — 247 PII checkpoints total.
python -c "from openmed import extract_pii; print([(e.label, e.text) for e in extract_pii('Dr. Pedro Almeida, CPF: 123.456.789-09, email: pedro@hospital.pt', lang='pt').entities])"Show per-language examples (Portuguese, Dutch, Hindi, Arabic, Japanese, Turkish)
from openmed import extract_pii
portuguese = extract_pii("Paciente: Pedro Almeida, CPF: 123.456.789-09, telefone: +351 912 345 678", lang="pt", use_smart_merging=True)
dutch = extract_pii("Patiënt: Eva de Vries, BSN: 123456782, telefoon: +31 6 12345678", lang="nl", use_smart_merging=True)
hindi = extract_pii("रोगी: अनीता शर्मा, फोन: +91 9876543210, पता: नई दिल्ली 110001", lang="hi", use_smart_merging=True)
arabic = extract_pii("المريضة ليلى حسن، الهاتف +20 10 1234 5678، الرقم القومي 29801011234567.", lang="ar", use_smart_merging=True)
japanese = extract_pii("患者 佐藤 花子、電話 +81 90 1234 5678、マイナンバー 1234 5678 9012.", lang="ja", use_smart_merging=True)
turkish = extract_pii("Hasta Ayşe Yılmaz, telefon +90 532 123 45 67, TCKN 10000000146.", lang="tr", use_smart_merging=True)
for r in (portuguese, dutch, hindi, arabic, japanese, turkish):
print([(e.label, e.text) for e in r.entities])A Docker-friendly FastAPI service with request validation, shared pipeline preload, and unified error envelopes.
pip install "openmed[hf,service]"
uvicorn openmed.service.app:app --host 0.0.0.0 --port 8080
# or with Docker
docker build -t openmed:1.5.5 .
docker run --rm -p 8080:8080 -e OPENMED_PROFILE=prod openmed:1.5.5curl -X POST http://127.0.0.1:8080/pii/extract \
-H "Content-Type: application/json" \
-d '{"text":"Paciente: Maria Garcia, DNI: 12345678Z","lang":"es"}'Model lifecycle (v1.5.5): free memory on demand with GET /models/loaded, POST /models/unload, and a keep_alive idle window:
OPENMED_SERVICE_KEEP_ALIVE=10m uvicorn openmed.service.app:app --host 0.0.0.0 --port 8080
curl -X POST http://127.0.0.1:8080/models/unload -H "Content-Type: application/json" -d '{"all":true}'See the full REST service guide.
Full guides at openmed.life/docs.
| Getting Started | Analyze Text | Model Registry |
| PII Detection Guide | Anonymization | Batch Processing |
| Configuration Profiles | REST Service | MLX Backend |
OpenMed's guardian is a fluffy Persian cat styled as a tiny Avicenna (Ibn Sina) — the great Persian physician whose Canon of Medicine was the world's standard medical text for some 600 years. He keeps watch over the open book of medical knowledge, in a palette built around Persian turquoise (fīrūza): a local-first guardian for your most private data.
Contributions welcome — bug reports, feature requests, and PRs alike.
- Open an issue
- Translations welcome — help complete the other-language READMEs linked in the switcher at the top.
OpenMed builds on excellent open-source work — particular thanks to OpenAI (the Privacy Filter architecture), NVIDIA (the Nemotron PII dataset), Hugging Face (transformers & the model ecosystem), Apple (MLX), and the Faker maintainers.
Released under the Apache-2.0 License.
@misc{panahi2025openmedneropensourcedomainadapted,
title={OpenMed NER: Open-Source, Domain-Adapted State-of-the-Art Transformers for Biomedical NER Across 12 Public Datasets},
author={Maziyar Panahi},
year={2025},
eprint={2508.01630},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2508.01630},
}If OpenMed is useful to you, a star helps others discover it.
Built by the OpenMed team
Website · Docs · X / Twitter · LinkedIn