How enrituraj/healthpulse works
Project Type & Purpose
• Type: Full-stack web application with predictive-analytics features.
• Purpose: Enable users to register/login, upload health data (activity, sleep, nutrition), store it in MongoDB, run ML models to predict diabetes/heart‐disease risk, and view personalized dashboards and reports. An admin panel allows user/route management.
Main Components
a. Frontend
• Browser UI: HTML/CSS/JS served by Flask templates (api/templates/…)
• Static assets: CSS, JS, images (api/static/…)
b. Backend (Flask API)
• Entry point: api/index.py
• Routes: Public (register/login/my_profile/health pages), Admin (dashboard, user/route CRUD)
• Controllers: Handle form submissions, file uploads, session management (SECRET_KEY), user auth
c. Data Storage
• MongoDB: stores users, health metrics, prediction results
• Connection via MONGO_URI (env var)
d. Machine Learning Module
• Pretrained models (diabetes, heart-disease) loaded in backend
• Data preprocessing & prediction logic invoked on health data submission
e. External/Deployment
• Hosting: Vercel config (vercel.json) or other cloud
• Environment variables (.env)
Relationships & Data Flow
• User Browser → HTTP(S) → Flask Routes
• Authenticated routes ↔ Session cookie
• Form uploads → Backend
• Prediction request → ML Module → Prediction results
• CRUD calls → MongoDB
• Admin Browser → Admin Blueprint/Routes → MongoDB
• Deployment platform pulls code → ENV setup → Flask app runs → connects to MongoDB Atlas
Architectural Patterns & Principles
• MVC (Flask “View”=templates, “Controller”=route handlers, “Model”=MongoDB schemas/collections)
• Single-process monolithic service (no microservices)
• Separation of concerns:
• Static vs template rendering
• Public vs admin routes
• Auth middleware
• Scalability: stateless Flask instances behind load balancer, MongoDB cluster
Key Technologies & Libraries
• Flask (web framework, routing, templating)
• Jinja2 (HTML templates)
• Flask-Login / session (authentication)
• PyMongo or Flask-Pymongo (MongoDB client)
• scikit-learn (or similar) for predictive models
• HTML/CSS/JavaScript (static UI)
• Vercel (deployment) or alternative PaaS
• Use meaningful labels: e.g., “User Service,” “Auth Middleware,” “Diabetes Model.”