makuachteny/medibridge

Telehealth chatbot designed to address critical healthcare accessibility challenges, particularly in underserved communities across African countries # MediBdridge AI MediBridge AI is a telehealth chatbot designed to address critical healthcare accessibility challenges, particularly in underserved communities across African countries. This AI-powered solution provides free preliminary medical consultations, bridging gaps in healthcare access for patients facing geographic and economic barriers to traditional medical services. ## Features - Medical Inquiry Handling: Responds to user questions with relevant medical advice. - Conversational AI: Provides human-like responses to simulate doctor-patient interactions. - Customizable: Easily retrainable with new datasets to improve accuracy and relevance. - Scalable: Built to handle multiple user queries efficiently. ## Dataset The chatbot is fine-tuned using the LinhDuong/chatdoctor-200k dataset, which contains 200,000 medical conversation pairs derived from the research paper ChatDoctor: A Medical Chat created using a pretrained GPT-2 Model Fine-Tuned using Medical Domain Knowledge. ### Dataset Features - Input: Contains conversational prompts from the users. - Instruction: Provides context for the model to understand the user's query and generate an appropriate response. - Output: Contains conversational responses of the model. Reference: arxiv.org ## Repository Structure ```bash MediBridgeAI/ ├── data/ │ ├── medical_conversations.csv # Dataset file containing medical Q&A pairs ├── src/ │ ├── doctor_ai.ipynb # Jupyter Notebook for data preprocessing, model training, and evaluation │ ├── model.py # Python script defining the chatbot model architecture │ ├── chatbot_interaction.py # Script for chatbot interaction (CLI-based) ├── models/ │ ├── saved_model/ # Directory containing the trained model in TensorFlow SavedModel format │ ├── model_weights.h5 # Optional: Saved model weights for quick loading ├── backend/ │ ├── app.py …