In the heart of data : machine learning applications for improved heart failure outcome prediction
View/
Keywords
Machine learning models ; Data mining ; Performance metrics ; Hyperparameter tunning ; Biomedical informatics ; Oversampling ; Undersampling ; Feature selection ; Principal component analysis ; Binary classification ; Μοντέλα μηχανικής μάθησης ; Εξόρυξη δεδομένων ; Μετρικές απόδοσης ; Βελτιστοποίηση υπερπαραμέτρων ; Βιοϊατρική πληροφορική ; Υπερδειγματοληψία ; Υποδειγματοληψία ; Επιλογή χαρακτηριστικών ; Ανάλυση κύριων συνιστωσών ; Δυαδική κατηγοριοποίησηAbstract
The aim of this dissertation is to investigate and evaluate the effectiveness of
various machine learning models (Random Forest, Decision Tree, Support Vector
Machine, Logistic Regression, k-Nearest Neighbors and Gradient Boosting) and
methods (Feature selection, Undersamlig, Oversampling and Principal component
analysis), in predicting the occurrence of death events in patients with heart failure. The
study employs a diverse set of models on a dataset comprising medical records of heart
failure patients, released by Ahmad et al. (2017). Through rigorous analysis,
hyperparameter optimization, and exploration of data preprocessing techniques, the
research seeks to develop a robust framework capable of accurately classifying heart
failure patients based on their risk of experiencing a death event. The findings aim to
contribute valuable insights to the evolving landscape of precision healthcare.
Specifically, by testing and subsequently selecting the most effective methods/models
of machine learning and addressing critical factors/errors present in the datasets, a more
efficient determination of results is achieved.
Postgraduate Studies Programme
Ψηφιακά Συστήματα και ΥπηρεσίεςDepartment
Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα Ψηφιακών ΣυστημάτωνNumber of pages
133Language
EnglishCollections
Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα