🩺 Diabetes Predictor AI

An advanced AI-powered diabetes risk assessment tool built with Streamlit and machine learning. This application uses sophisticated data preprocessing and multiple trained models to predict diabetes risk with high accuracy.

🌟 Features

🤖 Advanced AI Prediction

• Intelligent model fallback system for maximum reliability
• StandardScaler preprocessing for optimal accuracy
• Real-time confidence scoring

🎨 Modern UI/UX Design

• Beautiful gradient-based interface with custom CSS
• Responsive design that works on all devices
• Interactive animations and visual feedback
• Real-time risk assessment indicators
• Professional medical theme with intuitive icons

📊 Comprehensive Analytics

• Interactive probability visualization with Plotly
• Health profile radar charts comparing user vs normal ranges
• Real-time risk factor identification
• Detailed confidence metrics and recommendations

🔧 Robust Technical Implementation

• Intelligent error handling for model compatibility issues
• Support for multiple scikit-learn versions
• Automatic fallback model creation if needed
• Comprehensive input validation and preprocessing
• Professional logging and user feedback

🚀 Quick Start

Prerequisites

• Python 3.10 or higher
• pip package manager

1. Clone the Repository

git clone <your-repository-url>
cd M3_Final

2. Set Up Virtual Environment

# Create virtual environment
python -m venv kaggleenv

# Activate virtual environment
# On macOS/Linux:
source kaggleenv/bin/activate
# On Windows:
kaggleenv\Scripts\activate

3. Install Dependencies

pip install -r requirements.txt

4. Run the Application

streamlit run diabetes_predictor_app.py

5. Open in Browser

The application will automatically open in your default browser at:

http://localhost:8501

📁 Project Structure

M3_Final/
├── 📱 diabetes_predictor_app.py      # Main Streamlit application
├── 📊 01_data_cleaning.ipynb         # Data preprocessing notebook
├── 🤖 02_train_model.ipynb           # Model training notebook
├── 📈 diabetes.csv                   # Original dataset
├── ✨ diabetes_cleaned.csv           # Preprocessed dataset
├── 📦 requirements.txt               # Python dependencies
├── 🗂️ model/                         # Trained models directory
│   ├── diabetes_model.pkl            # Main trained model
│   ├── diabetes_model_quick_tuned.pkl # Quick-tuned variant
│   ├── diabetes_model_FAST_recall.pkl # Recall-optimized model
│   └── scaler.pkl                    # StandardScaler for preprocessing
├── 🐍 kaggleenv/                     # Virtual environment
└── 📖 README.md                      # This file

🧠 Technical Implementation

Machine Learning Pipeline

1. Data Preprocessing

• StandardScaler normalization for all numerical features
• Missing value handling with median imputation
• Outlier detection and treatment using IQR method
• Feature engineering for optimal model performance

2. Model Training

• Random Forest Classifier as primary model
• Hyperparameter tuning using GridSearchCV
• Cross-validation for robust performance evaluation
• Multiple model variants for different optimization goals

3. Features Used (8 Key Health Indicators)

• Pregnancies - Number of pregnancies
• Glucose - Plasma glucose concentration (mg/dL)
• BloodPressure - Diastolic blood pressure (mm Hg)
• SkinThickness - Triceps skin fold thickness (mm)
• Insulin - 2-Hour serum insulin (μU/mL)
• BMI - Body mass index (kg/m²)
• DiabetesPedigreeFunction - Genetic predisposition score
• Age - Age in years

Application Architecture

🎨 Frontend (Streamlit + Custom CSS)

• Modern gradient design with Inter font family
• Responsive layouts using Streamlit columns
• Custom animations (pulse, bounce effects)
• Interactive components with real-time feedback
• Professional medical styling with appropriate color schemes

🔧 Backend Logic

• Model loading with error handling for compatibility issues
• Intelligent fallback system creating dummy models if needed
• Real-time risk assessment with immediate visual feedback
• Data validation ensuring realistic input ranges
• Comprehensive error reporting with user-friendly messages

📊 Visualization Engine

• Plotly integration for interactive charts
• Real-time probability bars showing prediction confidence
• Radar charts comparing user profile vs normal ranges
• Risk indicator system with color-coded warnings
• Professional medical chart styling

🔬 Model Performance

Metric	Score
Accuracy	~81-85%
Precision	~80-84%
Recall	~75-85%
F1-Score	~78-82%
AUC-ROC	~0.85-0.90

🎯 Usage Guide

Basic Usage

1. Enter patient information in the input form
2. Review real-time risk indicators as you type
3. Click "PREDICT DIABETES RISK" for AI analysis
4. Review detailed results including confidence scores
5. Analyze visualizations for comprehensive understanding

Input Guidelines

• All fields are required for accurate prediction
• Use realistic medical values for best results
• Refer to normal ranges shown in tooltips
• The app handles data normalization automatically

Interpreting Results

• 🚨 High Risk: Immediate medical consultation recommended
• ⚠️ Medium Risk: Regular monitoring and lifestyle changes
• ✅ Low Risk: Continue healthy habits and routine checkups

🛠️ Development

Adding New Models

1. Train your model using the provided notebooks
2. Save as .pkl file in the model/ directory
3. Update the model loading logic in diabetes_predictor_app.py

Customizing UI

• Modify CSS in the st.markdown() sections
• Update color schemes in the gradient definitions
• Add new components using Streamlit's component library

Extending Features

• Add new health indicators in FEATURE_INFO
• Implement additional visualization types
• Enhance risk assessment algorithms

📋 Dependencies

Core Requirements

streamlit==1.28.0
pandas==2.0.3
numpy==1.26.4
scikit-learn==1.6.1
joblib==1.3.2
plotly==5.15.0

Development Tools

jupyter
kagglehub
pydeck
pillow

🚨 Important Disclaimers

⚠️ Medical Disclaimer: This application is designed for educational and screening purposes only. It should NOT be used as a substitute for professional medical advice, diagnosis, or treatment. Always consult qualified healthcare providers for medical decisions.

🔧 Technical Note: The application includes intelligent compatibility handling for different scikit-learn versions and will create fallback models if needed.

🤝 Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Pima Indians Diabetes Database for the training dataset
• Streamlit team for the excellent web framework
• Scikit-learn contributors for the machine learning tools
• Plotly team for the visualization library

📞 Support

If you encounter any issues or have questions:

1. Check the error messages - the app provides detailed troubleshooting
2. Verify your Python version and dependencies
3. Ensure all model files are present in the model/ directory
4. Check scikit-learn compatibility if you see model loading errors

---

🩺 Diabetes Predictor AI | Powered by Machine Learning & Advanced Data Processing

Built with ❤️ for better healthcare screening