Short communicationStatistical analysis of the 180 degree walking turn: Common patterns, repeatability and prediction bands of turn signals
Introduction
Turning is an essential movement which is required in nearly every daily activity task. Turning is a challenging task and there is evidence that turns are associated with a higher risk of falling [1]. Previous studies indicate that quantitative parameters of turning are useful markers that are sensitive to age effects and different pathologies. Namely, they were shown to differentiate healthy elderly from those with a mild cognitive impairment [2] or PD patients [3], and elderly adults with daily living disabilities [4]. A need to consider turning manoeuvres in routine clinical practice has been suggested [5].
Widely used data acquisition devices output continuous curves. These curves are expressed as a function of the time or percentage of turn. Usually, a single parameter is computed from the continuous curve. Extracted parameters are typically a minimal, maximal, mean value, or the value at a specific event (e.g. at heel strike). In comparison to single parameter analysis, analysis of the continuous curve is more informative [6].
In research and clinical practice, four types of issues related to curve analysis are encountered. The first issue concerns whether the curves of subject groups have the same pattern. This is important in selecting parameters for quantification of the turning manoeuvre because the quantification of curves with different patterns may not be adequate. Moreover, when the curves have a similar pattern then they can be processed to build the mean curve which serves as a representative curve for a given subject group.
The second issue concerns the repeatability of curves. When a turn task is performed repeatedly, e.g. within one session or an inter-session, the recorded curves should be characteristic for the subject to represent a usable parameter.
The third problem deals with classification of individual curves, i.e. deciding whether the subject's curve belongs to specific population (e.g. subject groups) or not. Statistical methods used in analysis of single parameters are not suitable for continuous curves. It has been shown that prediction bands are an adequate statistical tool when applied to continuous curves of gait [[7], [8], [9], [10]], cervical spine movement [11], and scapulo-humeral coordination [12]. Using prediction bands, the range of likelihood kinematics of subject groups can be defined. Determination of a “normal” range is necessary for researchers and clinicians to classify whether assessed curves belong to the same population as the training curves.
Finally, the fourth problem covers the comparison of subject groups. The comparison of the mean differences between subject groups can be useful to designate attractive parts which differentiate between the groups. Such differences might lay a foundation for further quantitative analysis.
An increasing amount of research involving the 180° walking turn [[13], [14], [15], [16], [17]] utilizes inertial measurement units to acquire data about the translational and/or rotational component of movement. However, the suitability, i.e. common pattern, of various turn signals for continuous analysis have not been explored yet. Furthermore, the intra-subject repeatability of curves has not been determined. Regarding the confidence bands of the walking turn manoeuvre, there was only one study that presented confidence bands for lower limbs kinematics [18] and kinetics [19] in typically developing children. It would be beneficial to researchers if a range of normal kinematics, namely angular velocity and acceleration, were established for the 180° turn manoeuvre.
Based on the highlighted problems, we address the following aims: (1) to identify which signals are suitable for walking turn analysis (2) to calculate prediction bands for walking turn signals, (3) provide reference curves for two subject groups, namely older adults and Parkinson disease patients, and (4) suggest promising turn segments for further assessment.
Section snippets
Participants and data acquisition
In the study we included 27 older healthy volunteers (24 males, 3 females), mean age 64.2 (SD 8.3) years without history of neurological disorder, and 24 mild treatment-naive Parkinson disease (PD) patients (15 males, 9 females), mean age 59.2 (SD 11.9) years. The study was approved by the Ethics Committee of the General University Hospital in Prague, Czech Republic, and therefore performed in accordance with the ethical standards established in the 1964 Declaration of Helsinki. Written,
Common curve patterns
The inter-subject reliability of all acceleration curves was poor in both subject groups (ρ < 0.50). For angular velocity signals the reliability varies from poor (ρ < 0.50) to good (0.90 > ρ > 0.75). Yaw angular velocity of both groups demonstrated good reliability (ρ = 0.79 for PD, ρ = 0.86 for CG). For details refer to Table 1.
Intra-subject repeatability
The intra-subject reliability of the yaw angular velocity varies from moderate (ρ > 0.50) to excellent (ρ > 0.90). Other angular velocity curves showed reliability
Discussion
The current study employed statistical methods to analyse walking turn curves from an instrumented Timed-Up and Go test. Using intra-class correlation this study provides assessment of within-group common patterns and intra-subject repeatability of turn angular velocity and acceleration. Next, the range of the likelihood of the turn kinematics of the subject group was determined via prediction bands. Additionally, this study provides a unique comparison of turning curves from different subject
Acknowledgements
This work was supported by Czech health research council (Ministry of Health of the Czech Republic, Czech Republic) Grant no. 16- 28119a "Analysis of movement disorders for the study of extrapyramidal diseases mechanism using motion capture camera systems", by The Czech Science Foundation (Czech Republic), grant No. 16- 07879S “REM sleep behavior disorder: predicting the risk of neurodegeneration”, and Charles University, Progres Q27.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References (26)
- et al.
Differences in axial segment reorientation during standing turns predict multiple falls in older adults
Gait Posture
(2012) - et al.
Video task analysis of turning during activities of daily living
Gait Posture
(2007) - et al.
Bootstrap prediction and confidence bands: a superior statistical method for analysis of gait data
Gait Posture
(1999) - et al.
Statistical tools for clinical gait analysis
Gait Posture
(2004) - et al.
Kinematic angular parameters in PD: reliability of joint angle curves and comparison with healthy subjects
Gait Posture
(2008) - et al.
Gait of normal Hong Kong Chinese children: the bootstrap approach
Hong Kong Physiother. J.
(2000) Bootstrap prediction bands for cervical spine intervertebral kinematics during in vivo three-dimensional head movements
J. Biomech.
(2015)- et al.
Prediction bands and intervals for the scapulo-humeral coordination based on the Bootstrap and two Gaussian methods
J. Biomech.
(2014) - et al.
Cognitive function impacts gait, functional mobility and falls in fragile X-associated tremor/ataxia syndrome
Gait Posture
(2018) - et al.
Spatio-temporal parameters and lower-limb kinematics of turning gait in typically developing children
Gait Posture
(2013)
Ground reaction forces and lower-limb joint kinetics of turning gait in typically developing children
J. Biomech.
A guideline of selecting and reporting intraclass correlation coefficients for reliability research
J. Chiropr. Med.
The reliability of the quantitative timed up and go test (QTUG) measured over five consecutive days under single and dual-task conditions in community dwelling older adults
Gait Posture
Cited by (3)
The timed up & go test sit-to-stand transition: Which signals measured by inertial sensors are a viable route for continuous analysis?
2021, Gait and PostureCitation Excerpt :Slightly stronger intra-class correlations in the CG than PD were observed in all statistically significant results. This is consistent with a previous study of walking turn signals in PD and a CG [8]. An analysis of the common signal pattern indicated that ωpitch had the most characteristic patterns for StS, which corresponds with intuitive processing of ωpitch in previous studies [3,10].
Beyond timing and step counting in 360° turning-in-place assessment: a scoping review
2024, BioMedical Engineering OnlineThe Reliability of Pig Gait Inertial Signals: A Pilot Study
2021, IFMBE Proceedings