How souhib-khalbous/heart-disease-classification-using-ecg-signals works
Project type and purpose
• This is a data‐science/ML pipeline for “Heart Disease Classification Using ECG Signals.”
• It ingests raw ECG records (PTB database), preprocesses and extracts features, then trains and evaluates both traditional ML and deep-learning models.
File structure
• All logic lives in Jupyter notebooks:
• Preprocessing_ECG.ipynb & Preprocessing_try2_ECG.ipynb – data cleaning, filtering, normalization
• ML_Models_ECG.ipynb – KNN, SVM, Decision Tree, Random Forest + SMOTE, PCA
• ConvNetQuacke_ECG.ipynb & Design_DL_Model_ECG.ipynb – DL architectures (1D CNN, LSTM/GRU, VAE)
• No explicit source directories yet; notebooks implicitly group functionality.
README for additional insights
• Data source: PTB Database (ECG records)
• Preprocessing: Standardization, FIR bandpass filtering, normalization
• Feature Extraction: Instantaneous Frequency via Hilbert transform
• ML: scikit-learn models with 10-fold CV, SMOTE for imbalance
• DL: TensorFlow/Keras models with oversampling, binary cross-entropy, Adam
• Evaluation metrics: accuracy, sensitivity, precision, F1-score
Diagramming recommendations
• Rectangles for processing modules (preprocessing, feature extraction, modeling)
• Parallelograms for outputs (charts, metrics reports)
• Clouds or octagons for external dependencies (PTB Database, imblearn, TensorFlow)
• Blue = Data ingestion/storage
• Green = Data processing
• Orange = Modeling
• Purple = Evaluation & Reporting