moDgass/Alzheimer-s-Detection

# Alzheimer's Disease Detection using Machine Learning ## Project Overview This project implements a machine learning model to detect and classify Alzheimer's disease stages using clinical data. The model leverages diagnostic information, cognitive scores, and biomarker data to provide accurate disease classification. ## Objective The goal of this project is to: - Build a predictive model for Alzheimer's disease classification - Process and merge multi-source clinical datasets - Standardize and encode features for machine learning - Train a Random Forest Classifier for accurate diagnosis prediction ## Dataset The project uses clinical data containing: - **Diagnosis Target**: Actual diagnoses of patients - **Cognitive Scores**: Cognitive assessment results - **Biomarkers & Data**: Clinical measurements and imaging data (FDG-PET, etc.) The data is merged on patient identifiers (RID) to create a comprehensive feature set. ## Disease Classification The model predicts three categories: - **CN**: Cognitively Normal - **LMCI**: Late Mild Cognitive Impairment - **AD**: Alzheimer's Disease ## Technologies & Libraries - **Python 3.x** - **pandas**: Data manipulation and merging - **numpy**: Numerical computations - **scikit-learn**: Machine learning algorithms - StandardScaler: Feature scaling - train_test_split: Data splitting - RandomForestClassifier: Classification model - LabelEncoder: Categorical encoding - **matplotlib**: Data visualization - **openpyxl**: Excel file handling ## Project Workflow 1. **Data Loading**: Load multiple sheets from Excel file 2. **Data Cleaning**: Remove missing values 3. **Data Integration**: Merge datasets on patient ID 4. **Feature Scaling**: Standardize numerical features 5. **Feature Engineering**: Encode categorical variables 6. **Model Training**: Train Random Forest Classifier 7. **Evaluation**: Assess model performance ## Key Features - Multi-sheet Excel data integration - Automatic handling of missing values - Standardized feature scaling - Categorical variable encoding - Train-test split (80-20) - Random Forest ensemble learning ## Usage # Load the notebook and run cells sequentially # Ensure the data file "CSI_7_MAL_2324_CW_resit_data.xlsx" is in the same directory # The notebook will: # 1. Load and merge clinical datasets # 2. Preprocess the data # 3. Train the model # 4. Generate predictions ## Files - `Alzheimer's Detection.ipynb` - Main Jupyter notebook with complete pipeline - `CSI_7_MAL_2324_CW_resit_data.xlsx` - Clinical dataset (required) ## Model Details - **Algorithm**: Random Forest Classifier - **Train-Test Split**: 80-20 - **Feature Scaling**: StandardScaler - **Encoding**: LabelEncoder for target variable ## Model Performance Metrics The model's performance can be evaluated using: - Accuracy - Precision & Recall - Confusion Matrix - Classification Report ## Learning Outcomes This project demonstrates: - Data preprocessing and integration - Feature engineering techniques - Model training and validation - Medical data analysis - Machine learning best practices ## Future Enhancements - Feature importance analysis - Hyperparameter tuning - Cross-validation implementation - Additional classification algorithms comparison - Model interpretability with SHAP values - Web interface for predictions ## License This project is part of my Machine Learning Projects collection. ## Author ## Mohamed Diaby Gassama **Note**: This project requires the clinical dataset to run. Ensure all dependencies are installed before execution.