Predicts the most relevant medical specialist based on Bangla health-related text. Built with machine learning, it provides accurate real-time predictions and will be improved with larger datasets in the future.
# 🩺 Specialist Prediction Model
This project is a **Bangla Specialist Prediction Model** powered by deep learning (TensorFlow/Keras).
It predicts the most relevant **medical specialist category** based on a given Bangla health-related sentence.
---
## 🚀 Features
- Accepts **Bangla health-related queries** as input
- Predicts the most suitable **medical specialist**
- Built with **TensorFlow / Keras** and fine-tuned ResNet model
- Deployed via **Flask + Gunicorn** for production-ready API
- Supports **real-time predictions**
---
## 🌟 Model Performance
- **Accuracy:** 0.95070 (~95%)
- Highly reliable predictions for common medical symptoms
- Model will be improved further with **larger datasets** in the future
---
## 🌐 Live Demo
🔗 Link: Specialist Prediction Web App
Example Input:
shebapabo.com হাতের তালু ও পায়ের তালুর ত্বক মোটা হয়ে গেছে এবং গভীর ফাটল দেখা দিয়েছে।
Example Output:
{
"input": "আমার হাতের তালু ও পায়ের তালুর ত্বক মোটা হয়ে গেছে এবং গভীর ফাটল দেখা দিয়েছে।",
"predicted_specialist": "Skin"
}
---
## 📖 How It Works
1. Input a **Bangla health sentence**
2. Text is **preprocessed** and tokenized
3. Model predicts the **most probable specialist**
4. Result is displayed instantly
---
## ⚙️ Tech Stack
- **Python 3.10+**
- **TensorFlow / Keras**
- **scikit-learn**
- **Flask**
- **Gunicorn**
- **HTML/CSS/JS** for frontend
---
## 🔮 Future Improvements
- Expand training dataset for **more rare diseases**
- Increase model **accuracy and reliability**
- Improve **response speed** for real-time predictions
- Add **multi-language support**
---
## 📜 License
This project is licensed under the **MIT License**.
---
## 👨💻 Author
Developed by **Md Abdullah Al Maruf**
Email •
GitHub •
LinkedIn