OpenMed-PII-SuperClinical-Small-44M-v1

What it does

This model performs token classification to identify 54 types of personal identifiers, including names, addresses, and medical record numbers, in English clinical and biomedical text. It reports an F1 score of 0.959 on a stratified 2,000-sample test set from NVIDIA Nemotron-PII.

Best for

• De-identifying clinical notes
• Redacting patient records
• Anonymizing research datasets

What it detects (54 labels)

• Account number
• Age
• API key
• Bank routing number
• Biometric identifier
• Blood type
• Certificate license number
• City
• Company name
• Coordinate
• Country
• County
• Credit debit card
• Customer ID
• Card CVV
• Date
• Date of birth
• Date time
• Device identifier
• Education level
• Email
• Employee ID
• Employment status
• Fax number
• First name
• Gender
• Health plan beneficiary number
• HTTP cookie
• IPv4 address
• IPv6 address
• Language
• Last name
• License plate
• MAC address
• Medical record number
• Occupation
• Password
• Phone number
• PIN
• Political view
• Postcode
• Race ethnicity
• Religious belief
• Sexuality
• Social security number
• State
• Street address
• SWIFT/BIC code
• Tax ID
• Time
• Unique ID
• URL
• Username
• Vehicle identifier

Quality

Use it

pip install transformers torch

from transformers import pipeline
ner = pipeline("token-classification", model="OpenMed/OpenMed-PII-SuperClinical-Small-44M-v1", aggregation_strategy="simple")
print(ner("Patient prescribed 10 mg aspirin for hypertension."))

pip install --upgrade "openmed[onnx-runtime]"

from openmed import OnnxModel
model = OnnxModel.from_pretrained("OpenMed/OpenMed-PII-SuperClinical-Small-44M-v1-onnx-android")
for entity in model("The patient started imatinib after the consultation."):
    print(entity.to_dict())

pip install "openmed[mlx]"

from openmed import analyze_text
from openmed.core.config import OpenMedConfig
result = analyze_text("Patient started imatinib in March.", model_name="OpenMed/OpenMed-PII-SuperClinical-Small-44M-v1", config=OpenMedConfig(backend="mlx"))
for e in result.entities:
    print(e.label, e.text)