Humans naturally select a point at which to transition from walking to running when gradually increasing locomotor speed. This point is known as the walk-to-run transition (WRT). The WRT is traditionally expressed in terms of speed and is known to occur within a close range of 2.1 m/s, which is an accepted heuristic (i.e., empirically based, rounded) threshold value. Very little research exists defining the WRT in terms of cadence (steps/min) despite the fact that spatial temporal aspects of gait underlying the WRT include this parameter. Preliminary evidence suggests that the WRT may be associated with a cadence of 140 steps/min in adults. This overlooked approach to identifying the WRT may be better than speed because of the simplicity and accessibility of recording cadence in both lab- and free-living settings. Wearable technologies can be used to determine cadence in real-time in a variety of settings, and could be used in the future to expand our current knowledge of the WRT. In turn, this knowledge could be used to inform training practices and/or rehabilitation of gait disorders. The purposes of this secondary analysis of an existing treadmill-based data set were to: (1) identify the optimal WRT cadence threshold, and (2) compare the accuracy of the cadence cutpoint to the previous WRT indicators identified in literature (i.e., speed and Froude number). This secondary analysis focused only on the data collected from the 28 participants (20 men, 8 women) whose protocol was terminated due to selecting to run during the treadmill portion of the larger CADENCE-Adults study. The CADENCE-Adults protocol consisted of a series of five-minute bouts beginning at 0.2 m/s and increasing in 0.2 m/s increments, with each bout followed by two minutes of standing rest. Participants could choose to walk or run each bout. The cadence of the bout during which the participants chose to run was considered the WTR cadence, and ROC analyses were performed to determine the optimal cadence cutpoint. Sensitivity, specificity and overall accuracy were calculated to compare the accuracy of the speed and Froude values from literature to the calculated cadence cutpoint. In addition, these analyses were expanded post hoc to also examine the accuracy of the previously proposed cadence cutpoint from the literature and the speed and Froude cutpoint identified from the dataset. Following analyses, three cadence cutpoints (134, 139, or 141 steps/min) were identified that shared equal overall accuracy (92.9%); therefore, there was no single optimal cutpoint. This also occurred for the speed cutpoints, where both 1.9 and 2.0 m/s shared overall accuracies of 78.6%. The optimal Froude cutpoint identified was 0.46 (82.0% overall accuracy). The rank-order overall accuracy of previously identified cutpoints were: a cadence of 140 steps/min (91.1%), Froude number of 0.5 (76.8%) and speed of 2.1 m/s (66.1%). Based on the identified optimal cadence cutpoints, a heuristic range of running cutpoints was recommended anchored on specificity vs. sensitivity preferences. For researchers interested in identifying episodes more likely to be running behavior (with the preference that very few episodes of walking behavior are mistakenly identified), it would be best to use 140 steps/min. However, if they want to be as inclusive as possible in identifying episodes of running behavior (and can tolerate more mistakenly identified episodes walking behavior), they could use 135 steps/min. When applied to this dataset, 96.0% (24/25) of the individuals who were ≥140 steps/min were running, but this decreased to 92.5% (25/27) with ≥135 steps/min. In conclusion, cadence clearly performed much better in terms of overall accuracy when compared to traditionally used WRT indicators of speed and Froude numbers. The recommended heuristics cadence cutpoint range can be used by researchers who want to evaluate the locomotor patterns of individuals when analyzing free-living step-defined data collected using wearable devices.
| Identifer | oai:union.ndltd.org:UMASS/oai:scholarworks.umass.edu:masters_theses_2-1934 |
| Date | 15 July 2020 |
| Creators | Chase, Colleen |
| Publisher | ScholarWorks@UMass Amherst |
| Source Sets | University of Massachusetts, Amherst |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | Masters Theses |
Page generated in 0.0059 seconds