Pearlyx

Pearlyx uses AI-driven voice analysis for early Parkinson’s detection. We aim to expand to neurological and respiratory diseases, making screening faster, easier, and more accessible.

Comment

Inspiration

Pearlyx was inspired by the critical need for early detection of neurological diseases like Parkinson’s. Studies show that 90% of Parkinson’s patients experience voice changes, often before motor symptoms appear. Our goal was to create a non-invasive, AI-driven tool to help individuals and healthcare professionals detect early warning signs through voice analysis.

What It Does

Pearlyx analyzes subtle voice variations using AI and machine learning to detect potential early signs of Parkinson’s. Users can:

  • Record or upload voice samples for instant analysis.
  • Receive AI-powered insights on vocal tremors, breath control, and speech irregularities.
  • Get a comprehensive report with multiple voice metrics.
  • Use AI chat support to learn more about Parkinson’s and related conditions.

How We Built It

Pearlyx is a full-stack web application utilizing:

  • Frontend: React + TypeScript, styled with TailwindCSS, and optimized with Vite.
  • Backend: Python Flask server with a custom machine learning model for voice analysis.
  • AI Chat Support: Integrated Gemini AI for interactive user assistance.
  • Audio Processing: Used pydub and FFmpeg to process voice recordings.
  • Security & Privacy: Implemented environment variables and secure data handling for user privacy.

Challenges We Ran Into

  • Data Availability: Finding diverse, high-quality voice datasets to train our AI model.
  • Model Optimization: Balancing accuracy and performance to reduce false positives/negatives.
  • Real-Time Processing: Ensuring smooth, fast audio analysis without delays.
  • User Accessibility: Designing an intuitive UI that simplifies complex AI-driven results.

Accomplishments That We're Proud Of

  • Successfully developed and deployed an AI-powered voice analysis tool.
  • Built a working machine learning model to detect voice irregularities.
  • Integrated real-time AI chat support to enhance user experience.
  • Created an intuitive and visually appealing UI using modern frontend tools.
  • Laid the groundwork for expanding detection beyond Parkinson’s to other neurological and respiratory diseases.

What We Learned

Throughout this project, we gained expertise in:

  • Machine Learning for Healthcare: Understanding the complexities of AI in medical applications.
  • Audio Signal Processing: Handling and analyzing voice data efficiently.
  • Frontend-Backend Integration: Ensuring smooth communication between React and Flask.
  • Data Ethics & Security: Protecting sensitive health-related data while maintaining usability.

What's Next for Pearlyx

We aim to enhance and expand Pearlyx with:

  • Improved ML Models: Training AI to detect Alzheimer’s, pneumonia, and other conditions.
  • Real-Time Tracking: Allowing users to monitor vocal changes over time.
  • Mobile App Development: Expanding Pearlyx to Android and iOS.
  • Enhanced Data Privacy: Implementing encrypted local storage for better security.
  • Clinical Collaboration: Partnering with healthcare professionals to validate and refine our model.

We’re excited to continue improving Pearlyx and making AI-powered voice analysis a key tool in early disease detection!

Built With

Share this project:

Updates