Regular Article

Scaling behavior in measured keystroke time series from patients with Parkinson’s disease

¹ Department of Physics, McGill University, H3A2T8 Montreal, Canada
² Montreal Institute for Learning Algorithms (MILA), H2S3H1 Montreal, Canada
³ Department of Physics, Ferdowsi University of Mashhad, P.O. Box 1436, Mashhad, Iran
⁴ Department of Epidemiology and Biostatistics, School of Public Health, Isfahan University of Medical Sciences, Isfahan, Iran
⁵ Department of Physics, Sharif University of Technology, Tehran 11155-9161, Iran
⁶ Institute of Physics and ForWind, Carl von Ossietzky University, 26111 Oldenburg, Germany

^a e-mail: mohammed.r.rahimi.tabar@uni-oldenburg.de

Received: 16 November 2019
Received in final form: 21 February 2020
Published online: 1 July 2020

Abstract

Parkinson has remained as one of the most difficult diseases to diagnose, as there are no biomarkers to be measured, and this requires one patient to do neurological and physical examinations. As Parkinson is a progressive disease, accurate detection of its symptoms is a crucial factor for therapeutic reasons. In this study, we perform Multifractal Detrended Fluctuation Analysis (MFDFA) on measured keystroke time series for three different categories of subjects: healthy, early-PD, and De-Novo patients. We have observed different scaling behavior in terms of multifractality of the measured time series, which can be used as a practical tool for diagnosis purposes. Additionally, the source of the multifractality has been studied which shows that in healthy and early-PD subjects, multifractality due to the long-range correlations is stronger than the influence of its probability distribution function (PDF) fatness, while in De-Novo patients, both shape of PDF and long-range correlations are contributing to observed multifractality.