Shelly - Your Stroke Recovery Companion

Inspiration

Stroke recovery is one of the most challenging journeys a person can face, often feeling isolating and overwhelming. Traditional rehabilitation apps lack the emotional intelligence and personalized companion that patients desperately need. We were inspired to create Shelly after learning that 70% of stroke survivors experience depression during recovery. We wanted to build something that combines cutting-edge AI with genuine empathy - a digital companion that feels like a caring friend who's always there to encourage, guide, and celebrate every small victory.

What it does

Shelly is an AI-powered turtle companion that provides comprehensive stroke recovery support through:

  • Conversational AI Companion: ElevenLabs' advanced AI creates natural, empathetic conversations tailored to each patient's emotional needs
  • Personalized Recovery Plans: Customized exercises and learning modules based on stroke type, mobility level, and recovery goals
  • Voice-Enabled Interactions: Shelly speaks with a warm, encouraging voice and responds to patient needs in real-time
  • Progress Tracking: Visual progress charts, milestone celebrations, and adaptive goal setting
  • Health Management: Medication reminders, vital sign logging, and symptom tracking
  • Emergency Support: Quick access to emergency contacts and medical team information
  • Emotional Wellness: Daily mood check-ins, meditation guidance, and motivational messages

How we built it

Frontend & Mobile App:

  • Expo (React Native) for cross-platform mobile development (Android, IOS, Web)
  • NativeWind (Tailwind CSS) for consistent, turtle-themed design system
  • Custom turtle character animations and mood-based interactions
  • Responsive design optimized for accessibility needs

AI & Voice Integration:

  • ElevenLabs Conversational AI for natural language processing and voice synthesis
  • Custom client tools allowing Shelly to suggest specific exercises and learning modules
  • Real-time audio playback with expo-audio for seamless voice interactions

Backend & Database:

  • Supabase for authentication, user profiles, and progress data
  • PostgreSQL database with custom migrations for patient data
  • Row-level security for healthcare data privacy compliance

Deployment:

  • Netlify for web deployment with optimized builds
  • Cross-platform compatibility for iOS, Android, and web browsers

Challenges we ran into

Dependency Issues: Fixing Expo dependency issues proved difficult in bolt and required much manual debugging

AI Context Management: How do we use a conversational AI with actionable healthcare guidance without overstepping medical boundaries, required much thought.

Accessibility Design: Creating an interface that works for users with varying motor skills, cognitive abilities, and visual impairments while maintaining an engaging, non-clinical feel.

Cross-Platform Consistency: Ensuring the turtle animations, voice playback, and touch interactions work seamlessly across iOS, Android, and web platforms. We mainly developed with Expo Go and many elements did not fully work on the web without intervention.

Accomplishments that we're proud of

  • Empathetic AI Integration: Successfully created a conversational AI that feels genuinely caring and supportive, not robotic or clinical.
  • Tailored Actions From Companion: Providing patient tailored actionable items such as buttons to exercises or learning modules directly in the conversation
  • Comprehensive Recovery Platform: Built a complete ecosystem covering physical, cognitive, and emotional aspects of stroke recovery
  • Real-Time Voice Interactions: Implemented seamless voice conversations with Shelly that work across all platforms
  • Beautiful, Therapeutic Design: Created a visually calming, turtle-themed interface that reduces anxiety and promotes healing

What we learned

Healthcare UX is Critical: Every design decision impacts someone's recovery journey. We learned to prioritize accessibility, emotional comfort, and medical safety over flashy features.

AI Requires Careful Boundaries: Conversational AI in healthcare needs careful prompt engineering to be supportive without providing medical advice, requiring extensive testing and refinement.

Voice Technology Challenges: Implementing cross-platform audio with proper state management and user controls is complex, especially for web applications with varying browser support.

Patient-Centered Design: Real stroke recovery needs go far beyond exercises - emotional support, medication management, and emergency preparedness are equally crucial.

Technical Integration Complexity: Combining multiple AI services, databases, authentication, and cross-platform deployment requires careful architecture planning and extensive testing.

What's next for Shelly - Your Stroke Recovery Companion

Clinical Partnerships: Collaborate with rehabilitation hospitals and stroke centers to integrate Shelly into formal recovery programs, gathering clinical data to improve outcomes.

Advanced AI Capabilities: Implement computer vision for exercise form correction, voice analysis for speech therapy progress, and predictive analytics for recovery milestones.

Family & Caregiver Features: Extend the platform to include family members and caregivers, with progress sharing, communication tools, and caregiver support resources.

Telehealth Integration: Connect with healthcare providers for virtual check-ins, data sharing, and coordinated care plans directly through Shelly.

Expanded Recovery Programs: Add support for other neurological conditions like traumatic brain injury, Parkinson's disease, and multiple sclerosis.

Research & Analytics: Partner with medical researchers to analyze anonymized recovery data, contributing to stroke recovery research and evidence-based treatment improvements.

Global Accessibility: Implement multi-language support and cultural adaptations to serve stroke survivors worldwide, with region-specific medical guidance and cultural sensitivity.

Built With

  • bolt
  • elevenlabs
  • expo.io
  • netlify
  • supabase
Share this project:

Updates