MedNER-TR — Turkish Medical NER

Turkish Medical Named Entity Recognition model and demo. Use it to extract medical entities (drug, disease, symptom, test, organ) from Turkish clinical-style text.

Links

• Demo (Space): https://huggingface.co/spaces/tugrulkaya/medner-tr-demo
• Model: https://huggingface.co/tugrulkaya/medner-tr

Supported Entities

• ILAC — Medications
• HASTALIK — Diseases
• SEMPTOM — Symptoms
• ORGAN — Organs
• TEST — Medical tests

Quick Start

pip install transformers torch

from transformers import pipeline

ner = pipeline(
    "token-classification",
    model="tugrulkaya/medner-tr",
    aggregation_strategy="simple",
)

text = "Hastaya Parol 500mg baslandi."
results = ner(text)
print(results)

Example Outputs (Illustrative)

• Parol → ILAC
• ates, oksuruk → SEMPTOM

Performance

Reported metrics:

• F1: 99.49%
• Precision: 99.49%
• Recall: 99.49%
• Accuracy: 99.76%

Notes / Limitations

• Model performance may vary depending on:
  • spelling mistakes / informal writing
  • domain shift (different hospitals / note styles)
  • uncommon abbreviations
• Always validate outputs in clinical workflows.

License

MIT

Citation

If you use this project in academic work, please cite:

@misc{mednertr,
  title   = {MedNER-TR: Turkish Medical Named Entity Recognition},
  author  = {Tuğrul Kaya},
  year    = {2026},
  url     = {https://github.com/mtkaya/medner-tr}
}

Contact

• GitHub: https://github.com/mtkaya
• Hugging Face: https://huggingface.co/tugrulkaya