University of Twente Student Theses

The semi-automatic classification of patients with RBD and PD based on a new automatic EM detector, manual sleep stage scorings, and machine learning methods

Weij, M.E.C. van der (2023) The semi-automatic classification of patients with RBD and PD based on a new automatic EM detector, manual sleep stage scorings, and machine learning methods.

Full text not available from this repository.

Full Text Status:	Access to this publication is restricted
Abstract:	Introduction: Parkinson’s disease (PD) is one of the most common neurological diseases in the world. It is a progressive neurodegenerative disorder (NDD) that dramatically reduces the quality of life of patients and their families. PD is often associated with Rapid Eye Movement (REM) Sleep behavior disorder (RBD). RBD is a sleep disorder characterized by dream enactment, such as limb movement and vocalizations. Definitive diagnoses for both diseases are currently based on manual data scoring by a sleep expert, which is subjective, tedious, costly, and challenging. Therefore, there is a need for a digital diagnostic method for both disorders. Research objective: This study aimed to establish the suitability of Eye Movement (EM) characteristics from electroOculoGram (EOG) measurements in a PolySomnoGraphy (PSG) dataset for the (semi-)automatic classification of patients with RBD and PD with and without RBD from healthy controls, using minimal manual dependency, measurement equipment, and machine learning methods. Methods: From the PSG dataset, the EOG channels were used to develop an automatic EM detector and extract 64 different characteristics for the classification algorithm. After careful consideration, three feature selection methods and a Random Forest (RF) classifier were chosen for the semi-automatic patient classification. Results: The best classification algorithm was found using a combination of a Maximum Relevance Minimal Redundancy (MRMR) algorithm for feature selection and the TreeBagger algorithm for patient classification. Area Under the Curve (AUC) values of 1, 1, 0.90 and 0.99 for the classification of healthy subjects, RBD, PD without RBD, and PD with RBD, respectively, were found. Furthermore, from the confusion matrices, it was found that all patients were classified correctly using this classification algorithm. Conclusion: Using two EOG channels as an input for the semi-automatic patient classification algorithm developed, patients suffering from RBD and PD with and without RBD can be accurately classified from healthy subjects.
Item Type:	Essay (Master)
Clients:	Danish Technical University, Kongens Lyngby, Denmark
Faculty:	TNW: Science and Technology
Subject:	30 exact sciences in general
Programme:	Biomedical Engineering MSc (66226)
Link to this item:	https://purl.utwente.nl/essays/94038
Export this item as:	BibTeX EndNote HTML Citation Reference Manager

Show download statistics for this publication

Repository Staff Only: item control page