Revealing the Generality of 1/f Noise Based Spectral Characteristics of Human Activity Across Different Datasets

2023 International Conference on Noise and Fluctuations (ICNF), Grenoble, France, 2023, pp. 1-4, Publisher: IEEE (Institute of Electrical and Electronics Engineers)

DOI: doi.org/10.1109/ICNF57520.2023.10472765

 

Bálint Maczák 1, Csenge G. Horváth 2, Gergely Vadai 1, Róbert Bódizs 2

1 Department of Technical Informatics, University of Szeged, Szeged, Hungary

2 Institute of Behavioural Sciences, Semmelweis University, Budapest, Hungary

 

Abstract

There is substantial evidence about the scale-independent nature of human dynamics. Within the field of actigraphy, numerous findings support this claim, such as the emergence of power-law distributions and the presence of 1/f noise in human locomotor activity. Recently, we discovered that the spectral characteristic of human activity – which includes 1/f noise at frequencies greater than the daily periodicity – is universal, as both the raw acceleration of the

wrist and the diverse kinds of activity signals (i.e., acceleration data compressed in varied ways) follows the same spectral nature. Here, we demonstrate that this spectral characteristic

persists for daily human activity in general by analysing datasets from various sources containing activity or acceleration signals of healthy, free-living subjects.

 

Keywords: 1/f noise, actigraphy, power law, scale-free nature, human dynamics