How yuvraj-dhepe/hackhpi2025-tone-down works
A React Native frontend (found under the ToneDown directory).
2. A Flask backend (under flask_backend) that serves as the API between the mobile app and the statistical modeling.
3. A statistical modeling module (located in the backend directory) which uses Bayesian hierarchical models (including a Stan model) for personalized intervention recommendations.
Main Components
• Frontend (React Native):
• ToneDown app as the mobile client.
• UI components such as screens, navigational components (Expo Router), and assets (images, fonts).
• Mention supporting modules like storage and utility files.
• Backend (Flask API Server):
• Flask backend (flask_backend directory) responsible for handling API requests.
• Components like routing (app.py), middleware (e.g., use of Flask-CORS), and data processing using libraries like Pandas.
• Statistical Modeling:
• Statistical module as a separate service/component.
• Files such as backend_wrapper.py, single_user_sampler.py, hierarchical_sampler.py, and the Stan model (hier_reg.stan).
• Integration with CmdStan for Bayesian inference.
Communication Flows
• Optionally, include feedback loops where the statistical model outputs updated recommendations that the Backend then communicates back to the Frontend.
Architectural Patterns
• That the project uses a clear separation of concerns:
• A decoupled mobile client (React Native, Expo) from the server (Flask).
• A clear data-processing and statistical modeling layer.
• Mention that data flows via RESTful API endpoints and that asynchronous data storage (using AsyncStorage and SecureStore on the frontend) is used.
Relevant Technologies
• Under the Frontend side, list “React Native, Expo Router, AsyncStorage, Secure Store.”
• For the Backend, add “Flask, Flask-CORS, Pandas, JSON.”
• For the Statistical Modeling, list “CmdStan, Stan modeling language, Python (wrappers), Bayesian Hierarchical Models.”