Deep Learning Models for Predicting Cognitive Impairment in Parkinson's Disease Detection
Loading...
Date
2024
Authors
Journal Title
2024 IEEE 3rd World Conference on Applied Intelligence and Computing, AIC 2024
Journal ISSN
Volume Title
Publisher
Institute of Electrical and Electronics Engineers Inc.
Abstract
A chronic neurodegenerative disorder affecting the motor system is called Parkinson's disease. Cell degeneration results from it over time as it advances slowly. This is one of the most prevalent diseases in society and is difficult to diagnose. The body experiences both motor and non-motor deficits (such as smell and speech) as a result of a dopamine cell shortage in the brain. Speech problems are common among Parkinson's sufferers, as is well documented. When compared to normal individuals, speech signals from Parkinson's sufferers show notable changes. This study proposes to use voice signals' acoustic properties to classify Parkinson's diseases using a deep learning-based approach. First, a genetic algorithm is used for the auditory characteristics in order to identify useful aspects. The ReliefF feature selection approach and the genetic algorithm's performance are also contrasted. The second phase involves feeding the chosen characteristics into the Convolutional Neural Network (CNN) architecture that has been created. An accuracy of 93.29% is attained without feature selection, whereas 97.62% is attained with feature selection. � 2024 IEEE.
Description
Keywords
Deep learning, Dopamine deficiency, Neurodegenerative disorder, Parkinson's disease, ReliefF algorithm