Arabic Medical Classification Model (AraBERTv2)
This model classifies Arabic medical questions and answers into one of 89 medical categories such as Cardiology, Neurology, Endocrinology, etc.
It is fine-tuned on a private dataset of 800,000 Arabic (question, answer, classification) pairs to help in building Arabic healthcare QA systems, virtual assistants, and medical chatbots.
Model Details
| Property | Value |
|---|---|
| Base model | aubmindlab/bert-base-arabertv2 |
| Language | Arabic (Modern Standard Arabic + some dialect) |
| Task | Text Classification |
| Number of classes | 89 |
| Training samples | ~800,000 |
| Evaluation metrics | Accuracy, F1 Macro, F1 Weighted |
| Hardware used | 1 × NVIDIA GPU (CUDA 12.8) |
| Framework | Transformers + PyTorch |
⚙️ How to Use
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
import torch.nn.functional as F
model_id = "YourUsername/arabic-medical-classifier-arabertv2"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForSequenceClassification.from_pretrained(model_id)
model.eval()
question = "ما هي أعراض ارتفاع ضغط الدم؟"
answer = "من أهم الأعراض الصداع والدوخة والنزيف الأنفي أحيانًا."
text = question + tokenizer.sep_token + answer
inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True, max_length=512)
with torch.no_grad():
outputs = model(**inputs)
probs = F.softmax(outputs.logits, dim=-1)
pred_id = torch.argmax(probs, dim=-1).item()
print("Predicted class:", model.config.id2label[pred_id])
print("Confidence:", probs[0][pred_id].item())