Diabetes Prediction
A machine-learning web project led during GSSoC, improved through contributor mentorship and testing.
https://diabetes-prediction-gamma.vercel.app/
Problem
Health-risk prediction tools can be hard to understand if the model result is shown without context. The project needed to collect user health inputs, produce a diabetes-risk prediction, and explain the surrounding patterns clearly enough for a general user.
What I Built
Diabetes Prediction combines a React front end with a Python/Flask machine-learning backend. Users can enter demographic and health data, receive a prediction, and review visualizations that explain relationships between risk factors.
The product includes a prediction page, a visualization page, and an about page that explains the project mission and data-security intent.
Technical Decisions
The front end uses React, JavaScript, HTML, CSS, and Tailwind CSS for the user experience. The backend uses Python and Flask, with pandas and scikit-learn for data processing and prediction. Matplotlib and Seaborn support the visualization layer.
During GirlScript Summer of Code 2024, I worked as Project Admin and Mentor. My focus was not only feature work, but contributor coordination, review quality, testing, and making the UI responsive.
Outcome
I led 30+ contributors, reviewed and merged 60+ pull requests, improved model accuracy by 15%, and added automated testing plus responsive UI improvements.