Recent advancements in machine learning have opened new avenues for preventing fatal ventricular arrhythmia by accurately measuring and analyzing QT intervals. This paper presents virtual sensor based on an expert system designed to prevent the risk of fatal ventricular arrhythmias associated with QT-prolonging treatments. The expert system categorizes patients into three risk levels based on their electrocardiogram-derived QT intervals and other clinical data, such as age or sex, facilitating informed decision-making and reducing the workload for healthcare professionals. Expert systems, known for their effectiveness in classifications with limited data, are particularly advantageous in this context. They not only achieve better standard metrics but also offer interpretability that other machine learning models lack. The proposed system’s performance has been rigorously compared against various machine learning algorithms, demonstrating superior efficiency as evidenced by confusion matrices, standard classification metrics, and receiver operation point curves. With an accuracy of 96.5%, the expert system proves to be the best option among the models evaluated, optimizing patient care and treatment outcomes by enabling more frequent and precise electrocardiogram assessments.
Authors
- Sebastián García-Galán,
- Jose A. Cabrera-Rodriguez,
- Francisco Javier Maldonado-Carrascosa,
- Nicolás Ruiz-Reyes,
- prof. dr hab. inż. Małgorzata Szczerska link open in new tab ,
- Pedro Vera-Candeas,
- Francisco D. Gonzalez-Martinez,
- Francisco J. Canadas-Quesada,
- Alfonso J. Cruz-Lendinez
Additional information
- DOI
- Digital Object Identifier link open in new tab 10.1016/j.bspc.2024.107255
- Category
- Publikacja w czasopiśmie
- Type
- artykuły w czasopismach
- Language
- angielski
- Publication year
- 2025
