Multimodal Prediction of Alzheimer’s Disease

This project implements a comprehensive multimodal approach for predicting Alzheimer’s Disease using machine learning and deep learning techniques. Developed as part of Washington University in St. Louis’s CSE 419A: Introduction to AI for Health course, the project utilizes the OASIS-1 dataset to analyze various imaging and clinical data for early detection and prediction of Alzheimer’s Disease.

Key Features

• Multimodal data integration from OASIS-1 dataset
• Deep learning models for image analysis
• Machine learning models for clinical data analysis
• Comprehensive performance evaluation metrics
• Visual analysis of model predictions
• Combined classifier leveraging both imaging and clinical data

Technologies & Tools

• Python 3.12
• Deep Learning Frameworks: TensorFlow/Keras, PyTorch
• Machine Learning: Scikit-learn, XGBoost
• Data Analysis: Pandas & NumPy
• Visualization: Matplotlib & Seaborn
• Development Environment: Jupyter Notebooks

Project Components

• Main implementation notebook with data processing, model training, and evaluation
• Detailed project report in NeurIPS format
• CNN architecture visualization and implementation
• Final classifier architecture combining multiple modalities
• Comprehensive performance analysis and visualization

Project Documentation

• GitHub Repository
• Project Report: Available in NeurIPS format
• Presentation: Final Demo slides covering methodology and results