Diabetes Prediction - ML Web Application
A web app that predicts diabetes risk from health metrics with clear charts explaining the results. Led 30+ open-source contributors during GSSoC 2024 to build it.
https://diabetes-prediction-gamma.vercel.app/
The Problem
Most machine learning projects just spit out a single prediction percentage, which leaves regular users confused about why they got that result and which specific health factors (like glucose, BMI, or age) caused it.
How It Solves It
This web application lets users enter simple health metrics and immediately calculates their risk. Crucially, it includes interactive charts that visually show how their numbers compare to healthy reference ranges, making the prediction easy to understand and discuss with a doctor.
How It's Built
- Python Flask Backend: Runs the trained scikit-learn machine learning classification pipeline.
- React Frontend: Clean, responsive form interface with input validation.
- Visual Analytics: Interactive graphs built with Seaborn and Matplotlib showing which risk factors contributed most to the prediction.
What It Means
I served as Project Admin and Mentor during GirlScript Summer of Code (GSSoC) 2024, guiding 30+ contributors, reviewing and merging 60+ pull requests, and improving overall prediction accuracy by 15% through community collaboration.