Master's Thesis - degree project in health informatics, Stockholm University and Karolinska Institutet

Managing highly dimensional electronic health record data for machine learning models: improving predictions of 30-day hospital readmissions and mortality

Data source: MIMIC III

Files 0-2: data pre-processing

File 3 random forest model with varying parameters of each dimensionality reduction technique

File 4: SVM model with varying parameters of each dimensionality reduction technique

File 5: neural network model (multi-layer perceptron) with varying parameters of each dimensionality reduction technique

File 6: evaluation metrics