AI-Based Disease Prediction System

Built a web-based symptom checker predicting over 40 diseases using multiple machine learning algorithms, including deep neural networks for cancer detection and traditional ML models for disease classification.

Overview

Developed as part of the “Software Project Management” course at Sichuan University, this comprehensive web-based system provides users with quick and accurate predictions for various diseases based on symptoms and medical data. The platform integrates multiple machine learning models to offer predictions across different disease categories.

Challenge

Creating a unified platform that combines traditional machine learning algorithms with deep learning models to provide accurate disease predictions across multiple medical domains, from symptom-based classification to image-based cancer detection.

Approach

Built a full-stack application with a React.js frontend and Django REST Framework backend, implementing multiple specialized prediction modules:

Disease Prediction Modules

  • Symptom-Based Disease Prediction: Implemented Gaussian Naive Bayes algorithm to predict over 40 diseases based on user-reported symptoms
  • Skin Cancer Detection: Developed Deep Neural Network model using TensorFlow/Keras for image-based skin cancer prediction
  • Lung Cancer Detection: Created convolutional neural network for lung cancer detection from medical images
  • Diabetes Prediction: Implemented Gradient Boosting Algorithm for diabetes risk assessment
  • Breast Cancer Detection: Utilized Random Forest Algorithm for breast cancer prediction based on medical data
  • Heart Disease Prediction: Deployed Support Vector Machine (SVM) model for cardiovascular risk assessment

System Architecture

  • Frontend: React.js with Material-UI and Bootstrap for responsive design
  • Backend: Django and Django REST Framework for API development
  • Authentication: Secure user authentication system for both patients and doctors
  • ML Integration: Seamless integration of Scikit-learn models and TensorFlow/Keras deep learning networks

Key Results

Successfully developed a multi-modal disease prediction system that provides accurate predictions across different medical domains, demonstrating practical application of machine learning in healthcare. The system supports both traditional ML algorithms and deep learning models, providing a comprehensive solution for various disease prediction scenarios.

Technologies

Python, Scikit-learn, Django, React.js, TensorFlow, Keras, REST APIs, Bootstrap, Material-UI