PerfectHealth

AI that understands your body — helping you fix issues before they become problems.

Comment

🧠 Inspiration

I noticed how often people push through fatigue without realizing their bodies are signaling burnout. Even with fitness bands tracking metrics like heart rate, sleep, and strain, most users don’t know what the data actually means. I wanted to bridge that gap — transforming raw sensor readings into early, actionable insights that warn users before short-term fatigue turns into long-term burnout or sickness.

⚙️ What it does

PerfectHealth is an AI-powered health companion that interprets wearable data and tells you what your body is trying to say.
It analyzes key physiological signals — recovery, strain, heart-rate variability, and sleep efficiency — to detect early warning patterns of exhaustion or illness.
The system then provides clear, personalized feedback such as:

“Your recovery has dropped 40% this week due to poor sleep. Reduce strain and aim for 9h rest to bounce back by Friday.”

By combining data analytics, AI reasoning, and visual dashboards, PerfectHealth helps users understand why their body feels off — and what to do about it before it’s too late.

🏗️ How I built it

  • Collected and structured daily metrics from fitness bands
  • Stored and processed data securely using AWS S3
  • Queried and explored datasets using AWS Athena
  • Built a serverless backend on AWS Lambda that calls Bedrock (Claude Sonnet 4.5) to generate AI health insights
  • Used Python (Pandas, Matplotlib, Scikit-learn) for data cleaning, trend detection, and feature engineering
  • Designed a responsive frontend dashboard (HTML/CSS/JavaScript) to visualize recovery, strain, and sleep correlations in real time

🚧 Challenges I ran into

  • Cleaning and aligning inconsistent data from multiple devices and export formats
  • Handling CORS and API Gateway configuration for the AWS Lambda frontend integration
  • Distinguishing between temporary fatigue and true burnout patterns in limited data
  • Managing time effectively while learning new AWS services within the hackathon window

🏆 Accomplishments I'm proud of

  • Built a fully functional AI insight pipeline using AWS Lambda + Bedrock within 24 hours
  • Created interpretable visualizations that reveal how recovery collapses before burnout
  • Delivered a clean web dashboard that fetches live AI-generated insights from AWS
  • Proved that wearable data can be transformed into early-warning systems for wellbeing

📚 What I learned

  • How to combine AWS S3 + Athena + Lambda for efficient, serverless data analysis
  • The critical role of data quality and context in any predictive health model
  • How even small, consistent habits (like an extra hour of sleep) can meaningfully shift recovery metrics
  • That AI becomes far more useful when it speaks the language of human health instead of numbers

🚀 What's next for PerfectHealth

  • Integrate with more fitness band APIs for continuous, real-time data
  • Expand the AI system to generate natural-language wellness summaries via Claude or GPT
  • Launch a mobile app for daily recovery notifications and progress tracking
  • Collaborate with wellness and corporate health platforms to provide proactive burnout alerts

Built With

Share this project:

Updates