MindTrack AI

Your AI Mental Health Companion

Comment

MindTrack AI: A Journey of Innovation in Mental Health Technology

Demo Credentials for judges

The Inspiration

Ever since elementary school, I've been driven by a simple yet powerful desire: to help people when they need it most. Growing up, I witnessed friends and family struggle with mental health challenges, often facing barriers to access professional support. The stigma, cost, and limited availability of mental health services created a gap that technology could potentially bridge.

The idea for MindTrack AI was born from this fundamental question: "What if we could create something that automatically helps people in need, providing 24/7 mental health support that's accessible, private, and genuinely helpful?"

I envisioned an AI-powered companion that could:

  • Track mood patterns and detect early warning signs
  • Provide personalized therapeutic interventions
  • Offer crisis support when traditional resources aren't available
  • Make mental health care as accessible as checking your phone

What I Built

MindTrack AI is a comprehensive mental health assistant that combines cutting-edge AI technology with evidence-based therapeutic approaches. The application features:

Core Features

  • Daily Mood Check-ins: Interactive mood slider (1-10 scale) with sleep quality, energy level, and stress tracking
  • AI-Powered Analytics: Smart pattern recognition that identifies mood trends and correlations
  • Crisis Detection System: Automated risk assessment with intervention protocols
  • Voice Therapy Sessions: AI-generated guided meditations and breathing exercises using ElevenLabs
  • Video AI Therapy: Interactive avatar therapist powered by Tavus for crisis intervention
  • Personalized Recommendations: Context-aware suggestions based on mood data and patterns

Technical Architecture

  • Frontend: React 18 + TypeScript with Tailwind CSS for a beautiful, responsive interface
  • Backend: Supabase database with comprehensive security policies
  • AI Integration:
    • Google Gemini AI for mood analysis and pattern recognition
    • ElevenLabs for natural voice generation
    • Tavus for video-based AI therapy sessions
    • Dappier for intelligent resource discovery
  • Real-time Features: Crisis detection, automated insights generation, and progress tracking
  • Multi-language Support: 9 languages for global accessibility

Advanced Capabilities

  • Smart Crisis Detection: Analyzes mood patterns, journal entries, and behavioral indicators
  • Personalized AI Insights: Weekly summaries and pattern analysis
  • Voice Emotion Recognition: Detects emotional cues from speech patterns
  • Secure Data Management: HIPAA-compliant encryption and privacy protection
  • Progressive Web App: Mobile-optimized experience with offline capabilities

How I Built It

The Bolt.new Experience

I chose Bolt.new as my primary development platform for several compelling reasons:

Why Bolt.new?

  • Rapid Prototyping: Transform ideas into functional applications within minutes
  • AI-Powered Development: Natural language descriptions become working code
  • Integrated Environment: Built-in IDE, testing, and deployment in one platform
  • Modern Framework Support: React, TypeScript, and modern tooling out of the box

The Development Process:

  1. Initial Concept: Started with a simple prompt: "Create an AI-powered mental health check-in app with mood tracking and crisis detection"

  2. Iterative Development: Built features incrementally:

    • Basic mood tracking interface
    • AI analysis integration
    • Voice session capabilities
    • Crisis detection algorithms
    • Video therapy features

  3. API Integration: Connected multiple AI services:

    // Example: Integrating Gemini AI for mood analysis
const analysis = await generateMoodAnalysis(entry, language);

  4. Database Design: Created comprehensive schema for:

    • User mood entries with metadata
    • AI insights and recommendations
    • Voice and therapy sessions
    • Crisis events and interventions

  5. UI/UX Refinement: Implemented:

    • Smooth animations with Framer Motion
    • Responsive design principles
    • Accessibility features (WCAG 2.1 AA)
    • Color psychology for calming effects

Technical Challenges Solved

Real-time Crisis Detection:

// Crisis detection algorithm
const performCrisisDetection = async (entries: MoodEntry[]) => {
  const riskFactors = analyzeRiskFactors(entries);
  const severity = calculateRiskSeverity(riskFactors);

  if (severity === 'high') {
    await triggerCrisisIntervention(userId, entry);
  }
};

AI-Powered Personalization:

// Personalized content generation
const insights = await generateAIInsights(
  entries,
  userPreferences,
  historicalData
);

What I Learned

Technical Discoveries

  1. AI Integration Complexity: Working with multiple AI APIs taught me about:

    • Token management and optimization
    • Rate limiting and error handling
    • Context preservation across sessions
    • Prompt engineering for consistent results

  2. Mental Health App Requirements: Developed understanding of:

    • HIPAA compliance and data security
    • Crisis intervention protocols
    • Evidence-based therapeutic approaches
    • Accessibility standards for vulnerable populations

  3. Real-time Data Processing: Learned to implement:

    • Efficient caching strategies
    • Background processing for AI analysis
    • Progressive enhancement for offline functionality

Development Insights

Bolt.new Advantages:

  • Speed: Reduced development time by 70% compared to traditional setup
  • AI Assistance: Smart code suggestions and error detection
  • Deployment: One-click deployment to production
  • Iteration: Rapid feature testing and refinement

Modern Web Development:

  • Component Architecture: Modular, reusable React components
  • Type Safety: TypeScript prevented numerous runtime errors
  • State Management: Efficient data flow with React hooks
  • Performance: Code splitting and lazy loading for optimal UX

Challenges Faced

1. Bolt.new Platform Limitations

Token Usage Constraints:

  • Challenge: Complex applications quickly consume available tokens
  • Impact: Had to optimize prompts and break down large features
  • Solution: Developed efficient prompt strategies and staged development

Project Size Limitations:

  • Challenge: Bolt.new has constraints on project complexity and file count
  • Impact: Couldn't implement all planned features within the platform
  • Solution: Hybrid approach - frontend in Bolt.new, backend separately

2. Backend Development Challenges

Why I Moved to VS Code: Due to Bolt.new's limitations, I had to manually implement the backend using VS Code:

-- Complex database schema for mental health data
CREATE TABLE mood_entries (
  id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
  user_id UUID REFERENCES auth.users(id),
  mood_score INTEGER CHECK (mood_score >= 1 AND mood_score <= 10),
  sleep_quality INTEGER CHECK (sleep_quality >= 1 AND sleep_quality <= 5),
  crisis_indicators TEXT[],
  created_at TIMESTAMP DEFAULT NOW()
);

Database Complexity:

  • Designed comprehensive schemas for mental health data
  • Implemented row-level security policies
  • Created complex analytics functions
  • Built crisis detection triggers

3. AI Integration Challenges

Multiple API Coordination:

// Coordinating multiple AI services
const generateComprehensiveAnalysis = async (entry: MoodEntry) => {
  const [moodAnalysis, crisisAssessment, recommendations] = 
    await Promise.all([
      geminiAI.analyzeMood(entry),
      performCrisisDetection([entry]),
      dappierAPI.getRecommendations(entry.mood_score)
    ]);

  return combineInsights(moodAnalysis, crisisAssessment, recommendations);
};

Challenges Overcome:

  • Token Optimization: Developed efficient prompt engineering
  • Error Handling: Implemented robust fallback mechanisms
  • Rate Limiting: Built intelligent caching and queuing systems
  • Context Management: Maintained conversation context across sessions

4. Mental Health App Considerations

Ethical Responsibilities:

  • Ensuring crisis detection accuracy
  • Maintaining user privacy and data security
  • Providing appropriate disclaimers about AI limitations
  • Implementing emergency intervention protocols

Technical Requirements:

  • HIPAA-compliant data handling
  • Real-time crisis detection
  • Multilingual support for global accessibility
  • Offline functionality for critical features

Impact and Future Vision

Current Achievements

Comprehensive Mental Health Platform:

  • Full mood tracking and analysis system
  • AI-powered crisis detection and intervention
  • Personalized therapy recommendations
  • Voice and video AI therapy sessions
  • Multi-language support (9 languages for now)

Technical Excellence:

  • Production-ready architecture
  • Secure data handling
  • Real-time AI processing
  • Mobile-optimized experience (not fully yet)

Future Enhancements

  1. Enhanced AI Capabilities:

    • Advanced pattern recognition
    • Predictive mental health analytics
    • Personalized intervention timing

  2. Professional Integration:

    • Therapist dashboard for monitoring
    • Clinical data export capabilities
    • Integration with healthcare systems

  3. Community Features:

    • Peer support networks
    • Anonymous community sharing
    • Group therapy sessions

Technical Stack Summary

{
  "frontend": {
    "framework": "React 18 + TypeScript",
    "styling": "Tailwind CSS",
    "animations": "Framer Motion",
    "forms": "React Hook Form",
    "charts": "Recharts"
  },
  "backend": {
    "database": "Supabase PostgreSQL",
    "authentication": "Supabase Auth",
    "storage": "Supabase Storage",
    "realtime": "Supabase Realtime"
  },
  "ai_services": {
    "analysis": "Google Gemini AI",
    "voice": "ElevenLabs",
    "video_therapy": "Tavus",
    "search": "Dappier"
  },
  "development": {
    "primary_platform": "Bolt.new",
    "additional_tools": "VS Code",
    "deployment": "Netlify",
    "version_control": "Git"
  }
}

Lessons for Future Developers

1. Choose the Right Tools for the Phase

Early Development (Bolt.new):

  • Perfect for rapid prototyping
  • Excellent for UI/UX iteration
  • Great for testing concepts quickly

Scaling Phase (Traditional Development):

  • More control over architecture
  • Better for complex backend logic
  • Necessary for production deployment

2. AI Development Best Practices

  • Start Simple: Begin with basic AI features before adding complexity
  • Token Management: Plan for token costs and optimization early
  • Error Handling: AI services can fail - always have fallbacks
  • User Experience: AI should enhance, not replace, human judgment

3. Mental Health App Development

  • Privacy First: User data security is non-negotiable
  • Crisis Protocols: Always provide clear paths to human help
  • Evidence-Based: Ground features in therapeutic research
  • Accessibility: Design for users in vulnerable states

Conclusion

Building MindTrack AI has been an incredible journey that combined my passion for helping others with cutting-edge technology. While I faced significant challenges with platform limitations and technical complexity, each obstacle taught me valuable lessons about modern web development, AI integration, and the unique requirements of mental health applications.

The project demonstrates that with the right combination of tools, persistence, and user-focused design, it's possible to create technology that can genuinely improve people's lives. MindTrack AI represents not just a technical achievement, but a step toward a future where mental health support is accessible to everyone, everywhere, at any time.

Most importantly, this project reinforced my belief that technology should serve humanity's most fundamental needs - and few needs are more fundamental than mental health and wellbeing.

"The best way to predict the future is to create it, and the best way to help people is to build tools that empower them to help themselves."

Built with: ❤️, Bolt.new, React, TypeScript, and a passion for mental health advocacy

Technical Implementation Highlights

Crisis Detection Algorithm:

const detectCrisisSignals = async (entries: MoodEntry[]) => {
  const recentEntries = entries.slice(0, 7);
  let riskScore = 0;

  // Analyze mood patterns
  const avgMood = recentEntries.reduce((sum, e) => 
    sum + e.mood_score, 0) / recentEntries.length;

  if (avgMood <= 2) riskScore += 40;

  // Check for crisis indicators in journal entries
  const crisisKeywords = ['hopeless', 'suicidal', 'self-harm'];
  const hasCrisisLanguage = recentEntries.some(entry =>
    crisisKeywords.some(keyword => 
      entry.journal_entry?.toLowerCase().includes(keyword)
    )
  );

  if (hasCrisisLanguage) riskScore += 50;

  return {
    riskLevel: riskScore >= 70 ? 'critical' : 
               riskScore >= 40 ? 'high' : 'low',
    recommendations: generateRecommendations(riskScore)
  };
};

AI-Powered Mood Analysis:

const generateMoodAnalysis = async (entry: MoodEntry, language: string) => {
  const prompt = `
    You are a compassionate mental health assistant. 
    Analyze this mood entry and provide supportive insights:

    Mood Score: ${entry.mood_score}/10
    Sleep Quality: ${entry.sleep_quality}/5
    Stress Level: ${entry.stress_level}/5
    Journal: ${entry.journal_entry}

    Provide a brief, supportive analysis in ${language}.
  `;

  const response = await geminiAI.generateContent(prompt);
  return response.text();
};

This project showcases the power of combining modern web development tools with AI services to create meaningful, impactful applications that can genuinely improve people's lives.

Built With

Share this project:

Updates