Is 15 minutes of human walking data the same as 1 hour?
Advisor Information
Vivien Marmelat
Location
Dr. C.C. and Mabel L. Criss Library
Presentation Type
Poster
Start Date
4-3-2016 12:45 PM
End Date
4-3-2016 2:15 PM
Abstract
Since healthy human movement exhibits complex patterns, it has been assumed that time series of walking data possess the property of scale invariance (SI), meaning the same information should exist for both short (200pts) and long (2000pts) time series. Detrended fluctuation analysis (DFA) characterizes SI and is commonly used to study the effects different movement disorders and walking conditions have on this these complex patterns. The purpose of this investigation was to determine whether this assumption of SI truly persists for stride time data of varying lengths from the same set of subjects. Footswitch data was collected from 10 healthy adults, who walked at a self-selected speed over-ground and on a treadmill. The stride-time data from two 1hr walking conditions were divided into shorter subsections and compared to each full time series using intra-class correlations, to determine how similar a subsection was to the full time series, and one-way repeated measures ANOVAs, to test whether the full time series were significantly influenced by sampling a subsection. From these statistical tests, we found SI persisted for shorter time series (<15min) in over-ground walking and still captures similar information for an individual subject. However, for treadmill time series less than 30min in length, SI deteriorates, especially for individual subject comparisons. These findings support using DFA for group comparisons, but demonstrate SI may not persist for shorter time series and depends upon the walking condition. Understanding these theoretical limitations of SI in walking has important implications for future study design when exploring SI.
Is 15 minutes of human walking data the same as 1 hour?
Dr. C.C. and Mabel L. Criss Library
Since healthy human movement exhibits complex patterns, it has been assumed that time series of walking data possess the property of scale invariance (SI), meaning the same information should exist for both short (200pts) and long (2000pts) time series. Detrended fluctuation analysis (DFA) characterizes SI and is commonly used to study the effects different movement disorders and walking conditions have on this these complex patterns. The purpose of this investigation was to determine whether this assumption of SI truly persists for stride time data of varying lengths from the same set of subjects. Footswitch data was collected from 10 healthy adults, who walked at a self-selected speed over-ground and on a treadmill. The stride-time data from two 1hr walking conditions were divided into shorter subsections and compared to each full time series using intra-class correlations, to determine how similar a subsection was to the full time series, and one-way repeated measures ANOVAs, to test whether the full time series were significantly influenced by sampling a subsection. From these statistical tests, we found SI persisted for shorter time series (<15min) in over-ground walking and still captures similar information for an individual subject. However, for treadmill time series less than 30min in length, SI deteriorates, especially for individual subject comparisons. These findings support using DFA for group comparisons, but demonstrate SI may not persist for shorter time series and depends upon the walking condition. Understanding these theoretical limitations of SI in walking has important implications for future study design when exploring SI.