How senthilcaesar/diabetes-nutrition-app works
Project type and purpose
• This is a Python-based, full-stack, AI-driven web application built with Streamlit.
• It collects user health metrics, socioeconomic data, optional genetic profiles, and uses OpenAI’s GPT-4 (and a RAG pipeline with Pinecone) to generate personalized nutrition plans for diabetes management.
• Key use cases: data collection pages, health & genetic analysis, AI content generation, retrieval-augmented Q&A from PDF resources.
File structure
• Root
• app.py: Streamlit entry point, page router
• Dockerfile / .devcontainer: containerization for consistent dev/test
• .streamlit/config.toml: Streamlit configuration
• apprunner.yaml: AWS App Runner deployment manifest
• app_pages/: UI pages (input_page, health_assessment_page, nutrition_plan_page, educational_resources_page, rag_qa_page)
• utils/: business logic & service layer (data_processing, llm_integration, genetic_processing, genetic_llm_integration, rag_system, pinecone_utils, UI components, visualizations)
• rag/: ingestion pipeline for PDF documents, Pinecone utility wrappers
• example_data/: sample genetic data to demo processing
• requirements.txt, README.md: dependencies & documentation
README for additional insights
• Describes modular separation: app.py (UI router), utils/ (services), app_pages/ (view layer), example_data/
• Lists core technologies: Streamlit UI, OpenAI GPT-4, Pinecone embeddings (via LangChain splitters), Matplotlib/Plotting, Pandas/Numpy, PyMuPDF for PDF extraction
• Mentions RAG Q&A system as an embedded feature leveraging retrieved PDF content
Final recommendations
• Every module in utils and app_pages as distinct nodes