OpenMed-PII-BioClinicalModern-Base-149M-v1

PII Detection Model | 149M Parameters | Open Source

Model Description

OpenMed-PII-BioClinicalModern-Base-149M-v1 is a transformer-based token classification model fine-tuned for Personally Identifiable Information (PII) detection in text. This model identifies and classifies 54 types of sensitive information including names, addresses, SSNs, medical record numbers, and more.

Key Features

• High Accuracy: Achieves strong F1 scores across diverse PII categories
• Comprehensive Coverage: Detects 50+ entity types spanning personal, financial, medical, and contact information
• Privacy-Focused: Designed for de-identification and compliance with HIPAA, GDPR, and other privacy regulations
• Production-Ready: Optimized for real-world text processing pipelines

Performance

Evaluated on a stratified 2,000-sample test set from NVIDIA Nemotron-PII:

Metric	Score
Micro F1	0.9509
Precision	0.9611
Recall	0.9409
Macro F1	0.9523
Weighted F1	0.9489
Accuracy	0.9932

Top 10 PII Models

Rank	Model	F1	Precision	Recall
1	OpenMed-PII-SuperClinical-Large-434M-v1	0.9608	0.9685	0.9532
2	OpenMed-PII-BigMed-Large-560M-v1	0.9604	0.9644	0.9565
3	OpenMed-PII-EuroMed-210M-v1	0.9600	0.9681	0.9521
4	OpenMed-PII-SnowflakeMed-568M-v1	0.9594	0.9640	0.9548
5	OpenMed-PII-SuperMedical-Large-355M-v1	0.9592	0.9632	0.9553
6	OpenMed-PII-ClinicalBGE-568M-v1	0.9587	0.9636	0.9538
7	OpenMed-PII-mClinicalE5-Large-560M-v1	0.9582	0.9631	0.9533
8	OpenMed-PII-ModernMed-Large-395M-v1	0.9579	0.9639	0.9520
9	OpenMed-PII-BioClinicalModern-Large-395M-v1	0.9579	0.9656	0.9502
10	OpenMed-PII-ClinicalE5-Large-335M-v1	0.9577	0.9604	0.9550

Best Performing Entities

Entity	F1	Precision	Recall	Support
biometric_identifier	0.998	0.996	1.000	228
credit_debit_card	0.998	0.995	1.000	213
race_ethnicity	0.997	1.000	0.995	193
blood_type	0.996	0.993	1.000	133
email	0.993	0.993	0.993	745

Challenging Entities

These entity types have lower performance and may benefit from additional post-processing:

Entity	F1	Precision	Recall	Support
unique_id	0.889	0.919	0.861	79
education_level	0.875	0.916	0.837	196
fax_number	0.856	0.786	0.939	98
time	0.848	0.886	0.813	460
occupation	0.602	0.704	0.526	688

Supported Entity Types

This model detects 54 PII entity types organized into categories:

Identifiers (16 types)

Entity	Description
account_number	Account Number
api_key	Api Key
bank_routing_number	Bank Routing Number
certificate_license_number	Certificate License Number
credit_debit_card	Credit Debit Card
cvv	Cvv
employee_id	Employee Id
health_plan_beneficiary_number	Health Plan Beneficiary Number
mac_address	Mac Address
medical_record_number	Medical Record Number
...	and 6 more

Personal Info (14 types)

Entity	Description
age	Age
biometric_identifier	Biometric Identifier
blood_type	Blood Type
date_of_birth	Date Of Birth
education_level	Education Level
first_name	First Name
last_name	Last Name
gender	Gender
language	Language
occupation	Occupation
...	and 4 more

Contact Info (4 types)

Entity	Description
email	Email
phone_number	Phone Number
fax_number	Fax Number
url	Url

Location (6 types)

Entity	Description
city	City
coordinate	Coordinate
country	Country
county	County
state	State
street_address	Street Address

Network Info (3 types)

Entity	Description
device_identifier	Device Identifier
ipv4	Ipv4
ipv6	Ipv6

Temporal (3 types)

Entity	Description
date	Date
date_time	Date Time
time	Time

Organization (1 types)

Entity	Description
company_name	Company Name

Usage

Quick Start

from transformers import pipeline

# Load the PII detection pipeline
ner = pipeline("ner", model="openmed/OpenMed-PII-BioClinicalModern-Base-149M-v1", aggregation_strategy="simple")

text = """
Patient John Smith (DOB: 03/15/1985, SSN: 123-45-6789) was seen today.
Contact: john.smith@email.com, Phone: (555) 123-4567.
Address: 456 Oak Street, Boston, MA 02108.
"""

entities = ner(text)
for entity in entities:
    print(f"{entity['entity_group']}: {entity['word']} (score: {entity['score']:.3f})")

De-identification Example

def redact_pii(text, entities, placeholder='[REDACTED]'):
    """Replace detected PII with placeholders."""
    # Sort entities by start position (descending) to preserve offsets
    sorted_entities = sorted(entities, key=lambda x: x['start'], reverse=True)
    redacted = text
    for ent in sorted_entities:
        redacted = redacted[:ent['start']] + f"[{ent['entity_group']}]" + redacted[ent['end']:]
    return redacted

# Apply de-identification
redacted_text = redact_pii(text, entities)
print(redacted_text)

Batch Processing

from transformers import AutoModelForTokenClassification, AutoTokenizer
import torch

model_name = "openmed/OpenMed-PII-BioClinicalModern-Base-149M-v1"
model = AutoModelForTokenClassification.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

texts = [
    "Contact Dr. Jane Doe at jane.doe@hospital.org",
    "Patient SSN: 987-65-4321, MRN: 12345678",
]

inputs = tokenizer(texts, return_tensors='pt', padding=True, truncation=True)
with torch.no_grad():
    outputs = model(**inputs)
    predictions = torch.argmax(outputs.logits, dim=-1)

Training Details

Dataset

• Source: NVIDIA Nemotron-PII
• Format: BIO-tagged token classification
• Labels: 106 total (53 entity types × 2 BIO tags + O)
• Splits: 50K train / 5K validation / 45K test

Training Configuration

• Max Sequence Length: 384 tokens
• Label Strategy: First token only (label_all_tokens=False)
• Framework: Hugging Face Transformers + Trainer API

Intended Use & Limitations

Intended Use

• De-identification: Automated redaction of PII in clinical notes, medical records, and documents
• Compliance: Supporting HIPAA, GDPR, and privacy regulation compliance
• Data Preprocessing: Preparing datasets for research by removing sensitive information
• Audit Support: Identifying PII in document collections

Limitations

⚠️ Important: This model is intended as an assistive tool, not a replacement for human review.

• False Negatives: Some PII may not be detected; always verify critical applications
• Context Sensitivity: Performance may vary with domain-specific terminology
• Challenging Categories: occupation, time, and sexuality have lower F1 scores
• Language: Primarily trained on English text

Citation

@misc{openmed-pii-2026,
  title = {OpenMed-PII-BioClinicalModern-Base-149M-v1: PII Detection Model},
  author = {OpenMed Science},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/openmed/OpenMed-PII-BioClinicalModern-Base-149M-v1}
}

Links

• Organization: OpenMed