Parkinson's disease (PD) is a progressive neurodegenerative disorder characterized by motor and non-motor symptoms. The diagnosis and management of PD have been significantly impacted by recent advancements in epidemiology, genetics, biomarkers, and therapeutic approaches.
This study aimed to identify potential risk factors for PD and assess their contribution to PD risk using the Explainable Boosting Machines (EBM) model, a machine learning approach. The dataset utilized in this research, accessible to researchers, comprised 2105 individuals and included 32 clinical and laboratory predictors across various categories, along with a response feature indicating PD diagnosis (yes/no). Statistical analyses, such as the Mann-Whitney U and Pearson chi-square tests, were conducted to determine significant differences between PD diagnosis groups.
The study identified several predictors as significantly different between the groups, including age, sleep quality, diabetes, depression, tremor, rigidity, bradykinesia, postural instability, and scores from assessment scales like UPDRS, MoCA, and Functional assessment. The EBM model effectively classified PD cases, demonstrating high accuracy, sensitivity, specificity, AUROC, and positive-negative predictive values. The "UPDRS" score emerged as the most influential predictor in the model, with higher scores indicating an increased risk of PD.
Future research, with more samples and predictors, can delve deeper into the interaction of these predictors and explore the potential for developing targeted interventions for PD prevention and management.
Key words: Parkinson’s disease, Risk factors, Explainable Boosting Machine
|
