LifeLens

Sign Up
Login
Home

Inspiration

We were inspired by the shortcomings of current AI doctor chatbots, which often rely only on patient self-reports. Miscommunication, incomplete information, and inaccurate self-diagnoses showed us the need for a more data-driven health assistant.

What it does

LifeLens combines:

5 ML Health Risk Models: Diabetes, High Blood Pressure, Allergy, Cancer, and Sleep Quality predictions
AI Health Assistant: GPT-4 powered conversations with voice interface
Healthcare Provider Discovery: Location-based clinic search with specialty matching
Real-time Health Data: Comprehensive profiles with trend analysis

How we built it

Frontend: React + TypeScript + Tailwind CSS for a modern, responsive interface.
Backend: Supabase for database, authentication, and real-time updates.
ML Models: Flask API serving models for diabetes, blood pressure, allergy, cancer, and sleep risk assessments.
AI Chat: OpenAI GPT integrated with voice recognition and text-to-speech.
Maps: Leaflet + Geolocation API for provider recommendations.

Challenges we ran into

ML Model Integration: Orchestrating 5 different models with varying data requirements
Voice Processing: Implementing reliable speech-to-text and text-to-speech with fallbacks
Performance: Optimizing real-time ML predictions for smooth user experience

Accomplishments that we're proud of

5 Production-Ready ML Models serving real-time health predictions
Complete Voice AI Pipeline with speech-to-text and text-to-speech
Healthcare Provider Integration with location-based specialty matching

What we learned

Machine Learning in Healthcare

Model Ensemble Methods: Combining multiple algorithms (XGBoost, LightGBM, Logistic Regression) for improved accuracy
Feature Engineering: Importance of preprocessing health data for optimal model performance
Real-time Inference: Challenges of serving ML models in production healthcare applications