An AI-powered medical chatbot that supports English, Hindi, Marathi, and Bengali.
It uses semantic search, automatic language detection, Google Translate, and confidence scoring to deliver accurate medical answers.
Developed by Manish Kumar
- π Supports 4 languages: English, Hindi, Marathi, Bengali
- π Semantic similarity search using SentenceTransformer
- π§ Automatic language detection with fallback logic
- π Automatic translation (question and answer)
- π Shows confidence score (0β1)
- π Displays answer source
- π Optional view of raw QA data
- πΌοΈ Clean, interactive UI built with Streamlit
Advanced_Multilingual_Medical_Chatbot/
β
βββ main.py # Streamlit app interface
βββ train_model.ipynb # Model training and evaluation
βββ langchain_helper.py # (optional future: helper functions for LangChain or modular logic)
βββ fixed_qa_dataset.json # Preprocessed Q&A data from MedQuAD
βββ trained_classifier.pkl # Saved Logistic Regression model
βββ embedded_qa_dataset.pkl # Embedded Q&A dataset using SentenceTransformer
βββ requirements.txt # Required Python packages
βββ README.md # Project documentation