synthetic-multi-med-notes-ner-gliner_multi-v2.1

A fine-tuned GLiNER v2.1 model for Named Entity Recognition (NER) in medical notes, trained on multilingual synthetic data.

Model Details

• Base Model: urchade/gliner_multi-v2.1
• Training Dataset: ErikCalcina/synthetic-multi-med-notes-ner-dataset-v1
• Task: Named Entity Recognition (NER)
• Languages: English, Italian, Spanish, German, French, Dutch, Greek, Portuguese, Slovenian

Trained Label Set

• Comorbidity
• Condition
• Date
• Device
• Drug
• Drug dose
• Event
• Measurement
• Observation
• Operation
• Procedure
• Rehabilitation
• Specimen
• Symptom
• Test
• Test score
• Treatment
• Treatment complication
• Visit

Usage

from gliner import GLiNER

model = GLiNER.from_pretrained("ErikCalcina/synthetic-multi-med-notes-ner-gliner_multi-v2.1")

text = (
    "On 2026-03-15, the patient visited cardiology with chest pain and fatigue."
    "ECG and troponin test were ordered. BP 150/95 mmHg, HbA1c 8.2%."
    "Diagnosed with hypertension and type 2 diabetes with obesity as comorbidity."
    "Started metformin 500 mg twice daily and amlodipine 5 mg daily."
    "Planned cardiac catheterization procedure and referral to rehabilitation."
)

labels = [
    "Comorbidity", "Condition", "Date", "Device", "Drug", "Drug dose", "Event",
    "Measurement", "Observation", "Operation", "Procedure", "Rehabilitation",
    "Specimen", "Symptom", "Test", "Test score", "Treatment",
    "Treatment complication", "Visit"
]

entities = model.predict_entities(text, labels, threshold=0.5)
for entity in entities:
    print(f"{entity['text']} -> {entity['label']} ({entity['score']:.3f})")

Training Details

• Synthetic Data: Multilingual medical notes generated from templates
• Training Quality: High-quality synthetic annotations for improved generalization

License

Licensed under Apache 2.0

Citation

If you use this model, please cite:

@model{synthetic_med_ner_gliner_2026,
  title={synthetic-multi-med-notes-ner-gliner_multi-v2.1},
  author={ErikCalcina},
  year={2026}
}