flan-t5-healthcare-bot

🤖 flan-t5-healthcare-bot

This model is a fine-tuned version of Google's t5-base model, specifically adapted for doctor-patient question answering in the healthcare domain.

The model was trained using a custom dataset derived from the ai-medical-chatbot.csv file, containing over 220,000 medical queries and answers between patients and doctors.

---

🧠 Model Architecture

Base model: t5-base
Fine-tuned for: Question Answering
Framework: transformers, PyTorch
Interface: Streamlit app (local deployment)

---

📊 Training Info

• Dataset: ai-medical-chatbot.csv
• Columns used:
  • Description = Question
  • Patient = Context
  • Doctor = Answer
• Tokenization: T5Tokenizer
• Epochs: 5
• Batch Size: 8
• Optimizer: AdamW

---

🧪 Example Inference

from transformers import T5Tokenizer, T5ForConditionalGeneration

tokenizer = T5Tokenizer.from_pretrained("omerdasc/flan-t5-healthcare-bot")
model = T5ForConditionalGeneration.from_pretrained("omerdasc/flan-t5-healthcare-bot")

question = "What are the symptoms of anemia?"
context = "Hi doctor, I feel tired and have pale skin. I read that these might be signs of anemia..."

input_text = f"question: {question} context: {context}"
inputs = tokenizer(input_text, return_tensors="pt", truncation=True, max_length=512)

output_ids = model.generate(**inputs)
print(tokenizer.decode(output_ids[0], skip_special_tokens=True))