GluChart

Blood glucose level prediction for patients with diabetes using machine learning.

Inspiration

Diabetes is a chronic condition that impairs insulin production, the hormone that regulates blood sugar. Patients with Type I diabetes must self-administer insulin, often guessing dosages before meals. Under‑ or overdosing can lead to dangerous highs or lows. We built GluChart to help our friend Aidan—and anyone adjusting to diabetes—by predicting their glucose levels ahead of time, improving dosing accuracy and quality of life.

How to Use

  1. Select Glucose, Meal, or Insulin.
  2. For Meal or Insulin, enter the value (grams of carbs or insulin units) and click Predict to view a 30‑minute “what if” forecast.
  3. To save an input for future predictions, click Submit; your glucose levels for the next 30 minutes will be charted.

What We Learned & Challenges

Brendan Leung

My first hackathon taught me frontend fundamentals and React useState across components.

Sriram Magesh

I mastered React.js and JS libraries; resolving Git merge conflicts was a key challenge.

Tim Stewart

Venturing into ML/AI was new; deploying on Azure was smoother than expected.

Jun Min Kim

I learned full‑stack development with Node.js and ML deployment—bold leaps of faith paid off.

How We Built It

GluChart is a full‑stack app hosted on Microsoft Azure. The ML model is built with Python, Keras, and scikit‑learn. The frontend is React and the backend uses Express.js.

Built With

  • Azure
  • React
  • Express.js
  • Chart.js
  • Node.js
  • Python
  • Keras & scikit‑learn
  • Figma

Check It Out

Awards & Recognition

Submitted to HackDavis 2023

  • Winner: Best Hack for Social Good
  • Winner: Best Healthcare Hack

Contributors

  • Jun Min Kim — Backend integration & ML model optimization
  • Brendan Leung — Frontend React components & animations
  • Tim Stewart — Machine learning model & Azure deployment
  • Sriram Magesh — Frontend design system