About the Project: SilentCare

🌟 Inspiration

The idea for SilentCare came from observing how stress, poor sleep, and excessive phone usage silently affect students and remote workers. Many existing health apps rely on wearables, cameras, or intrusive data collection, which raises privacy concerns.

I wanted to build a privacy-first AI solution that can detect early signs of burnout or sleep disruption without asking for personal messages, camera, microphone, or location data. The goal was to give actionable insights while maintaining full user privacy.

🛠 How We Built It

SilentCare is a dashboard + AI system that tracks daily phone behavior patterns like:

• Typing speed
• Screen time
• Night usage
  

Tech Stack

• AI Layer: Gemini + Genkit
  
  • Generates Health Drift Score and personalized recommendations
  • Processes multi-day trends and predictive insights
• Frontend/UI: NextJS + TypeScript + Tailwind CSS
  
  • Interactive dashboard with color-coded scores and line charts
• Optional Backend: Firebase Firestore
  
  • Stores user history for multi-day trend visualization

Health Drift Score Formula

We calculate the Health Drift Score (HDS) using a weighted combination of user behavior metrics:

[ HDS = w_1 \cdot f(\text{typing speed}) + w_2 \cdot f(\text{screen time}) + w_3 \cdot f(\text{night usage}) ]

Where (f(x)) normalizes each metric to a 0–100 scale, and (w_1 + w_2 + w_3 = 1).

🧠 What We Learned

• Integrating AI models: Learned how to use Gemini + Genkit to generate actionable insights and handle JSON outputs for frontend use.
• Frontend visualization: Built dynamic charts and dashboards with Tailwind and NextJS.
• Privacy-first design: Learned how to design AI systems that work on-device without collecting sensitive personal data.
• Hackathon workflow: Coordinated mock data simulation, predictive analysis, and real-time dashboard updates.

⚡ Challenges Faced

1. Simulating meaningful data without real phone metrics
   
   • Solved by creating mock datasets with realistic typing speed, screen time, and night usage trends
2. AI integration with frontend
   
   • Gemini + Genkit outputs required careful JSON handling and mapping to UI elements
3. Multi-day trend analysis
   
   • Needed to generate predictive Health Drift scores and dynamic charts
4. Privacy-first design
   
   • Ensuring no personal data leaks while still delivering actionable insights

🎯 Impact

SilentCare empowers users to:

• Track stress and sleep patterns
  
• Receive personalized, actionable tips
  
• Build healthy phone habits without compromising privacy
  

It’s a student-level, ethical, and innovative solution that demonstrates the power of AI in wellbeing, ready to inspire further development in the space.

Built With

• genai
• tailwindcss
• typescript