A two-stage variable selection and classification approach for Parkinson’s disease detection by using voice recording replications

• Published in: Computer methods and programs in biomedicine Vol. 142; pp. 147 - 156
• Main Authors: Naranjo, Lizbeth, Pérez, Carlos J., Martín, Jacinto, Campos-Roca, Yolanda
• Format: Journal Article
• Language: English
• Published: Ireland Elsevier B.V 01.04.2017
• Subjects: Algorithms; Artificial Intelligence; Bayes Theorem; Bayesian binary regression; Databases, Factual; Diagnosis, Computer-Assisted; Gibbs sampling; Humans; Internal Medicine; Models, Statistical; Other; Parkinson Disease - diagnosis; Parkinson’s disease; Regression Analysis; Replicated measurements; Reproducibility of Results; Sample Size; Sensitivity and Specificity; Software; Speech Acoustics; Variable selection; Voice; Voice features; Voice features; Replicated measurements; Gibbs sampling; Bayesian binary regression; Parkinson’s disease; Variable selection

•The motivating problem is the discrimination of people with PD from healthy subjects.•A two-stage variable selection and classification approach is developed.•The approach considers intra-subject variability in a proper way.•A Gibbs sampling-based method is derived to solve the computational proble...