Medical Question Category Classifier (DistilBERT)
Model Description
This model is a fine-tuned version of distilbert-base-uncased trained to classify medical and health-related questions into 20 distinct medical specialties (e.g., Cardiology, Pediatrics, Psychiatry, Surgery).
Intended Use
- Primary Use Case: Triage systems and automatic tag generation for medical Q&A platforms.
- Safety Notice: This model is designed strictly for categorization and routing purposes. It does not provide medical diagnosis or treatment advice.
Dataset Information
- Dataset:
openlifescienceai/medmcqa - License: Apache 2.0
- Classes (20): Anaesthesia, Anatomy, Biochemistry, Dental, ENT, Forensic Medicine, Gynaecology & Obstetrics, Medicine, Microbiology, Ophthalmology, Orthopaedics, Pathology, Pediatrics, Pharmacology, Physiology, Psychiatry, Radiology, Skin, Social & Preventive Medicine, Surgery.
Training Hyperparameters
- Base Model:
distilbert-base-uncased - Epochs: 3
- Batch Size: 16
- Learning Rate: 2e-5
- Optimizer: AdamW
Evaluation Metrics
- Accuracy: ~82.4%
- Weighted F1-Score: ~0.821
How to Use
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
model_name = "mahmoodulhassan23/medical-question-classifier"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)
question = "Child presenting with high grade fever, rash, and cough."
inputs = tokenizer(question, return_tensors="pt")
with torch.no_grad():
outputs = model(**inputs)
pred_class = torch.argmax(outputs.logits, dim=-1).item()
print(f"Predicted Class ID: {pred_class}")