Enhancing Parkinson’s Disease Detection Accuracy Through Vocal Biomarkers: A Refined Model Approach

International Research Journal of Engineering and Technology (IRJET)

e-ISSN: 2395-0056

Volume: 11 Issue: 04 | Apr 2024

p-ISSN: 2395-0072

www.irjet.net

Enhancing Parkinson’s Disease Detection Accuracy Through Vocal Biomarkers: A Refined Model Approach K. Nagaprakash1, P. Vandana2,M. Saikrishna3, P. Hari Manikanta4, N. Gowtham Naidu5 1Professor, Department of ECE, Seshadri Rao Gudlavalleru Engineering College, Gudlavalleru

2,3,4,5Student, Department of ECE, Seshadri Rao Gudlavalleru Engineering College, Gudlavalleru

---------------------------------------------------------------------***---------------------------------------------------------------------

Abstract - Parkinson's disease (PD) diagnosis and

In recent years, the utilization of machine learning (ML) methods has surged within medical research, presenting robust capabilities to bolster diagnostic precision and prognostic forecasting in various healthcare domains. By leveraging large-scale datasets containing diverse patient information, ML models can extract meaningful patterns and associations to aid in disease classification and prediction.

prognosis are pivotal for effective patient management. In this study, we explore the utility of a genetic algorithm (GA) for feature selection to enhance machine learning models applied to PD datasets. Our methodology encompasses dataset preprocessing to handle missing values and ensure data integrity. Subsequently, a GA-driven feature selection process is employed to identify salient features for classification tasks. Utilizing various classifiers, including Decision Trees (DT), Random Forest (RF), Logistic Regression (LR), AdaBoost, K Nearest Neighbors (KNN), and Support Vector Machines (SVM), we evaluate model performance with and without feature selection. Performance metrics such as accuracy are employed for rigorous evaluation. Our results demonstrate that models trained with feature selection consistently outperform those without. Notably, Decision Tree and Gradient Boosting classifiers achieve peak accuracies of 97% with feature selection, while maintaining accuracies of 92% without. Logistic Regression and Linear SVM exhibit slightly lower accuracies of 89% and 87%, respectively, without feature selection. These findings underscore the significance of feature selection in optimizing model accuracy for PD diagnosis and prognosis. In summary, our investigation underscores the effectiveness of genetic algorithm-based feature selection in enhancing machine learning models for the analysis of medical data, with particular emphasis on research related to Parkinson's disease (PD).

In this paper, we present a study aimed at improving PD diagnosis and prognosis through the application of ML techniques, specifically focusing on the utilization of a genetic algorithm (GA) for feature selection. Effective feature selection is pivotal in the development of machine learning models, as it discerns the most relevant features from a dataset, thereby trimming down dimensionality and computational intricacies. By selecting relevant features, ML models can achieve better generalization and performance on unseen data, thereby enhancing their utility in clinical settings. Our study encompasses several key components. Firstly, we preprocess a dataset comprising various clinical and demographic features related to PD patients, ensuring data quality and consistency. Subsequently, we employ a feature selection method powered by genetic algorithms (GA) to pinpoint the most discriminative features for classification tasks. Following this, we proceed to train and assess an assortment of machine learning classifiers, including DT, RF, LR, AdaBoost, KNN, and SVM leveraging both the identified features and the entire feature set.

Key Words: PD, Machine learning(ML),SVM,LR ,DT, Gradient boosting, KNN.

Through rigorous evaluation and comparison, we demonstrate the effectiveness of GA-driven feature selection in optimizing ML models for PD diagnosis and prognosis. Our findings underline how choosing the right features can make our computer models better at predicting and understanding Parkinson's disease. This helps doctors and researchers use computer programs more effectively in studying the disease and caring for patients with Parkinson's.

1. INTRODUCTION Parkinsonian syndrome manifests as a degenerative neurological condition, progressively impairing motor functions like tremors, bradykinesia, and muscle rigidity. Additionally, it presents a spectrum of non-motor symptoms that impact cognitive abilities, mood regulation, and autonomic processes. As the second most prevalent neurodegenerative disorder worldwide, PD poses significant challenges to healthcare systems and necessitates accurate diagnosis and prognosis for optimal patient care.

© 2024, IRJET

|

Impact Factor value: 8.226

In conclusion, our study adds to the expanding field of research dedicated to utilizing machine learning methods to improve medical diagnosis and personalized healthcare, especially in managing Parkinson's disease.

|

ISO 9001:2008 Certified Journal

|

Page 1092