HealthAI

HealthAI — Your complete wellness tracker with instant AI brain tumor screening.

Comment

🌟 Inspiration

HealthAI was inspired by the need for better accessible preventive health tools. Having seen firsthand how challenging it can be to manage ongoing symptoms and medications within the family, we envisioned a personal health journal (MedLog) anyone can use. Further, understanding the delays and barriers in brain tumor diagnosis, we aimed to empower users with TumorSense—a feature designed for quick, AI-based checks on MRI scans. Our mission: bridge practical daily health tracking with the power of AI insights—conveniently, securely, and all in one place.

🛠️ How We Built It

  • Frontend:

    • Mobile App: Built in Thunkable for rapid prototyping and accessibility across devices.
    • Web App: Developed in React + Vite, offering streamlined interfaces for MedLog and TumorSense.

  • Backend:

    • Prediction API: A Flask REST API serves our custom TensorFlow-based convolutional neural network model, exported as TFLite for fast, efficient inference.
    • Storage: MRI images handled via Cloudinary; personal health data and authentication managed with Firebase Firestore and Firebase Auth.

  • AI/ML:

    • Model: We designed, trained, and deployed a convolutional neural network on grayscale brain MRI images—carefully optimizing for fast, real-time use on the cloud.
    • Deployment: Our API is hosted on Render (free plan), which spins down after inactivity for cost efficiency.

  • Teamwork:

    • Developed collaboratively by two students, sharing roles from model development to full-stack integration and documentation.

👩💻 What We Learned

  • Data Security: Implemented practical encryption and access controls using Firebase and secure image flows with Cloudinary.
  • Building & Deploying Custom ML Models: Developed from scratch, then adapted for real-world inference and robust API service.
  • User Experience: Emphasized clear communication, straightforward onboarding, and reliable user authentication—critical for any health app.
  • Cloud Challenges: Navigated free-tier hosting quirks, including service spin-downs that impact initial response times.
  • Balancing AI Impact: Ensured our model is fast and accessible for everyday users, while recognizing its role as an informational aid—not a full medical solution.

🚧 Challenges We Faced

  • Model Optimization: Converted and tuned our own TensorFlow model until it was efficient enough for cloud use.
  • API Responsiveness: Managed user experience around occasional delays due to Render’s free tier spin-down; performance is fast once active.
  • Privacy: Set up strict permissions for sensitive health and MRI data using Firebase’s rules system.
  • Keeping it Simple: Designed intuitive forms for both medication and MRI uploads, avoiding jargon or overwhelming flows.

ℹ️ A Note on TumorSense

TumorSense is built for instant, AI-powered analysis of brain MRI images as a tool for informational and educational purposes. While it can provide valuable insights and assist with personal health awareness, it is not a substitute for medical advice, clinical diagnosis, or professional radiologist evaluation. Results should be seen as casual indications, not definitive assessments.

If you ever have concerns about any health or imaging results, please consult a qualified healthcare professional.

✨ The Journey

“HealthAI taught us that innovation lies at the intersection of empathy, data privacy, and accessible technology—building not just tools, but trust for users in their everyday wellness journey.”

🚀 What's Next for HealthAI

  • Expand TumorSense’s Dataset & Capabilities: Train the AI on a larger, high-quality, and more diverse set of brain MRI scans to improve performance and reliability. Explore expansion to detect other medical conditions through imaging.
  • Smart Health Analytics: Add visual trends, insights, and automated reminders in MedLog to help users understand and act on their health data.
  • Always-On Hosting: Upgrade the backend API to a more robust cloud service to eliminate startup delays after periods of inactivity.
  • Mobile App Distribution: Prepare for publication on Google Play and the Apple App Store so more users can benefit.
  • Professional Collaborations: Partner with clinicians and radiologists to validate AI accuracy and improve user workflow.
  • User Feedback & Community: Launch onboarding tutorials and direct feedback channels to iterate based on real-world user experience.
  • Enhanced Security & Privacy: Continually audit and upgrade privacy measures, ensuring compliance with new health data standards.
  • Share Feature:
    • QR Code Generation: Allow users to generate a QR code linking to their recent health summary or TumorSense results, making it easy for doctors to access up-to-date health data with a quick scan (while keeping strict privacy controls).
    • Direct, Secure Sharing: Add options for sharing reports via encrypted messages or links directly with healthcare providers—streamlining the process for referrals or consultations.

HealthAI continues to strive for a future where personal wellness management and AI-powered health insights are easy, secure, and accessible to all.

Built With

Share this project:

Updates