GENERATIVE USER INTERFACE
HEALTHCARE PREDICTION APPLICATION

Author: Dario Dang
Date: May 31, 2024
Code File

Project Overview

The Healthcare Prediction App is an interactive Streamlit-based tool that allows users to estimate their risk of stroke, diabetes, or heart attack based on personal health metrics such as age, BMI, and glucose levels. Users can input their data through a user-friendly form, view dynamic visualizations, and receive instant predictions powered by a trained Random Forest model. The app also provides guidance, average comparisons, and health recommendations for high-risk outcomes — making it practical for personal use or healthcare screening.

⚠️ This is a Demo Application – Not Production Ready

This application is a prototype demo designed to showcase the concept of health risk prediction using machine learning techniques. It currently supports predictions for stroke, diabetes, and heart attack based on user input, using models trained on publicly available datasets.

Because the app deals with healthcare-related outcomes, any real-world deployment must ensure high accuracy, data security, and ethical oversight. As such, several enhancements are required before production use:

  • Improve model accuracy through better datasets and hyperparameter tuning
  • Implement continuous retraining with real-time feedback
  • Add stricter validation logic for medical-grade input handling
  • Ensure compliance with health data privacy standards
  • Develop a mobile-optimized layout and accessibility support
  • Introduce a user feedback system to improve usability and trust

This version is for demonstration purposes only. Do not rely on it for clinical or medical decision-making.

Watch the demo to see how the app running locally step-by-step

THANK YOU FOR WATCHING!

← Back to Projects