AutiCare: Real-Time Stress Monitoring for Children with ASD
Fullstack Project - Software Engineering Degree
A real-time web application designed to help caregivers monitor physiological indicators of stress in children with ASD. Collects heart rate, temperature, and galvanic skin response (GSR) from wearable devices, processes data with a Random Forest ML model, and delivers live insights, historical trends, and instant notifications via a responsive dashboard.
Next.jsReactTailwind CSSFastAPISQLAlchemyPostgreSQLWebSocketsJWTAPSchedulerRandom Forest MLFramer Motion
Impact & Results
- Provided real-time stress detection with actionable recommendations for caregivers
- Enabled historical data analysis for long-term pattern recognition
- Integrated live sensor streaming with zero page refresh
- Demonstrated end-to-end full-stack development with ML integration
Architecture
- Frontend: Next.js + React with Tailwind CSS and Framer Motion for responsive, animated UI
- Backend: FastAPI with async SQLAlchemy and WebSocket support
- Database: PostgreSQL (async) for sensor readings, predictions, and user data
- Real-time: WebSockets for live sensor data and notification streaming
- ML: Pre-trained Random Forest model processing aggregated data every 5 minutes
- Scheduling: APScheduler for periodic stress prediction jobs
- Authentication: JWT with refresh tokens for secure sessions
Challenges
- Streaming high-frequency sensor data reliably over WebSockets
- Integrating and calibrating wearable sensor input in real-world conditions
- Balancing real-time performance with ML prediction accuracy
- Designing an intuitive UI for non-technical caregivers
Solutions
- Implemented robust WebSocket connection management with automatic reconnection
- Used simulated and real sensor data during development with fallback mechanisms
- Optimized prediction pipeline with periodic batch processing
- Iterative UI design with focus on clarity and immediate feedback
Key Takeaways
Full lifecycle of a real-time full-stack application with IoT integration
Effective use of WebSockets for bidirectional live communication
Integration of machine learning models into production backend services
Importance of user-centered design in health-related applications
Project Gallery
Browse through project illustrations
