Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Human Gait Analysis Metric for Gait Retraining

Human Gait Analysis Metric for Gait Retraining Hindawi Applied Bionics and Biomechanics Volume 2019, Article ID 1286864, 8 pages https://doi.org/10.1155/2019/1286864 Research Article Tyagi Ramakrishnan, Seok Hun Kim, and Kyle B. Reed University of South Florida, USA Correspondence should be addressed to Kyle B. Reed; kylereed@usf.edu Received 19 April 2019; Revised 25 July 2019; Accepted 10 September 2019; Published 11 November 2019 Guest Editor: Michelle Johnson Copyright © 2019 Tyagi Ramakrishnan et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The combined gait asymmetry metric (CGAM) provides a method to synthesize human gait motion. The metric is weighted to balance each parameter’seffect by normalizing the data so all parameters are more equally weighted. It is designed to combine spatial, temporal, kinematic, and kinetic gait parameter asymmetries. It can also combine subsets of the different gait parameters to provide a more thorough analysis. The single number quantifying gait could assist robotic rehabilitation methods to optimize the resulting gait patterns. CGAM will help define quantitative thresholds for achievable balanced overall gait asymmetry. The study presented here compares the combined gait parameters with clinical measures such as timed up and go (TUG), six-minute walk test (6MWT), and gait velocity. The comparisons are made on gait data collected on individuals with stroke before and after twelve sessions of rehabilitation. Step length, step time, and swing time showed a strong correlation to CGAM, but the double limb support asymmetry has nearly no correlation with CGAM and ground reaction force asymmetry has a weak correlation. The CGAM scores were moderately correlated with TUG and strongly correlated to 6MWT and gait velocity. 1. Introduction changes to the individual’s gait pattern from baseline to after the clinical intervention. Figure 1 shows an example of how a combined metric Researchers traditionally analyze a small set of gait parame- would be useful in analyzing an asymmetric gait pattern. ters in order to evaluate the outcomes of their techniques. This often leads to an overreliance on a few parameters and Many existing rehabilitation therapies can change different sets of gait parameters, but some make one parameter worse a focus on improving one gait parameter. Few studies in the while correcting others. Even in unimpaired walking, perfect gait literature aim to correct many gait parameters at the symmetry is not expected [1], so there is space for some same time. This traditional narrow approach lacks broader parameters to be asymmetric while the overall gait is within understanding of the interaction between various gait param- a reasonable bound. The CGAM distance (shown in orange eters and limits potential approaches that can lead to whole- in Figure 1) generates a single representation of the measured some rehabilitation techniques. In this research study, we gait parameters that generally scales with the global deviation examine our combined gait asymmetry metric (CGAM) to give a representation of the overall gait pattern. We use from symmetry. The deviation of each measure is scaled based on the variance within that measure, so measures that stroke for examining this combined metric because it affects generally have larger magnitudes of asymmetry (e.g., forces) several different aspects of an individual’s gait, and many of will be scaled so that each gait parameter has a similar influ- these aspects are asymmetric. Although we focus on mea- ence on the overall metric. If a therapy reduces the CGAM sures of asymmetry, this combined method is not limited by the type or number of parameters evaluated. Our hypoth- distance, the overall gait has improved even though some of the individual parameters might have gotten worse. Without esis is that the outcomes of the combined metric will partially a combined metric, it is difficult to determine whether the correlate to functional clinical outcome measures. We also gait is improving when looking at individual gait parameters. use this combined metric to determine if there have been 2 Applied Bionics and Biomechanics analysis like principle component analysis (PCA) and singu- Rehabilitated Asymmetric gait pattern 1 lar variable decomposition (SVD) to reduce dimensionality gait pattern to make the data computation easier [11]. The processed data is then classified using the Euclidean or similar distances [11]. These distances become the scores which form the cen- tral part of the gait metric. Another study by Hoerzer et al. [9] proposed the comprehensive asymmetry index (CAI) which Rehabilitated combined gait asymmetry using PCA and Euclidean dis- Rehabilitated gait pattern 3 tances. CAI was effective in identifying that running with gait pattern 2 shoes reduces gait asymmetry compared to barefoot running. Gait parameter 2, A prior study used a combination of Mahalanobis distances e.g., step time asymmetry with data reduction techniques on a preprocessed dataset to analyze kinematic and kinetic gait parameters [8]. They Perfect symmetry developed several metrics to classify the data and showed that they can successfully classify the abnormal data from a standard normal dataset. The precursor to CGAM used Typical able-body gait (slightly asymmetric) a symmetry index processed using PCA measured using Mahalanobis distances. Without the restrictions of dimen- sionality reduction, CGAM served as a versatile gait asymme- Figure 1: Representation of the multidimensional gait parameter space. The orange lines represent the distance each gait is from a try metric [12–14]. symmetric gait (CGAM distance), which helps determine how far away a gait is from ideal. CGAM can also aid in ascertaining 1.3. Effects of Stroke on Gait and Rehabilitation. The analysis whether the overall gait pattern is improving (even if some of the in this paper uses an existing dataset from an experimental parameters are getting worse). CGAM can incorporate more stroke therapy to examine the effects of combining and dimensions than the three shown, but that is hard to visualize. jointly assessing gait as opposed to individually assessing a single parameter. We focus on individuals with stroke 1.1. Gait Measurements. Gait data is typically collected using because they inherently have different capabilities on each motion capture, force plates, and/or wearable sensors. Many side and are asymmetric; as such, it is unlikely that they can ever regain complete symmetry in all parameters. However, variables portray various facets of human gait. There are spa- tial parameters such as step length defined by the distance it may be possible to achieve a balanced gait where some covered from the heel strike of one foot to the heel strike of parameters are slightly asymmetric, but none of them are the opposite foot. There are temporal parameters such as step excessively large. Our proposed joint metric helps to balance time defined as the time taken between opposite heel strikes. all of the parameters. We examine before and after the therapy to help understand what changes have occurred. Then, there is swing time, which is the time taken from toe- off to heel strike of the same foot. Double limb support is Gait after stroke becomes asymmetric (or hemiparetic) as the time spent when both legs are on the ground. The termi- a consequence of altered neuromuscular signals affecting leg nal double limb support is used for this research study. There motor areas, typically hyperextension at the knee and are kinematic parameters associated with joint angles of the reduced flexion at the hip, knee, and ankle [15–17]. Hemi- paretic gait is characterized by significant asymmetry in ankle, knee, and hip joints. Hip joints in the case of individ- uals with stroke and amputees also show abduction and temporal (e.g., time spent in double limb support) and adduction. The kinetic parameters include vertical ground spatial (e.g., step length) measures of interlimb coordination reaction forces, propulsive or push-off forces during toe-off, [15, 18, 19]. Propulsive force of the paretic limb is reduced braking forces during initial contact or heel strike, and ankle, compared to the nonparetic limb, as are work and power of the paretic plantar flexors [19, 20]. The significant decrease knee, and hip joint moments. Further, some of these param- eters are more easily identified by sight alone (e.g., step in propulsive force results in smaller overall step lengths, length, cadence, and gait velocity) while others are nearly which in turn affects the patient’s gait velocity. Finally, verti- impossible to quantify without a sensor (e.g., forces and joint cal ground reaction forces (GRFs) are decreased on the moments) [2]. paretic limb relative to the nonparetic limb [21], reflecting diminished weight bearing and balancing capabilities by the 1.2. Gait Metrics. Several gait metrics combining multiple paretic limb. gait parameters have been used clinically to evaluate different Some of the rehabilitation techniques used to restore gait gait impairments. These metrics can also be used to classify impaired by stroke involve some form of asymmetric pertur- gait based on different types of information. There are two bations that try to restore the symmetry between the paretic and nonparetic sides [22]. Split-belt treadmills are one types: qualitative [3, 4] and quantitative [5–7] metrics. Many metrics rely on either kinetic or kinematic data to categorize method to apply this rehabilitation technique. The split-belt different gait motions and behaviors. Some metrics have the treadmill has two treads that can move at different velocities, ability to jointly analyze kinetic and kinematic parameters which are used to exaggerate the asymmetry of the individ- [8, 9]. Machine learning has been used to classify and differ- ual. When the tread speeds are made the same after training, the subject typically has some after-effects that are more entiate gait patterns [10]. Most gait metrics use statistical CGAM metric Gait parameter 3, e.g., push-off force Gait parameter 1, e.g., step length Applied Bionics and Biomechanics 3 that are multiplied across the dataset in the numerator. To symmetric than when they started [23]. The after-effects are usually improved spatial and temporal symmetry. Unfortu- balance the influence of the inverse of covariance, it is divided nately, these after-effects only partially transfer to walking by the sum of the inverse covariance matrix, equation (2). on the ground. There are other rehabilitation techniques This change to the formulation makes the modified CGAM such as body-weight support [24], robotic [25], functional represent the scores closer to the percent asymmetry while electrical stimulation [26], transcranial magnetic stimulation still serving as a combined measure of all the gait parameter [27], and full-body gait exoskeletons [28]. Each of the tech- asymmetries. niques have their merits and train the individual in a special- ized manner, which means a combination of these methods 3. Methods may provide additional benefits to the person. The analysis performed in this paper used data collected as part of a separate clinical study. The novel shoe tested was 2. CGAM Derivation designed to improve the overall gait symmetry and gait func- The metric presented here has the potential to help categorize tion of an individual poststroke. The efficacy of the device is and differentiate between multiple asymmetric gaits [29]. discussed in another paper [31]. That study data is used here CGAM is based on Mahalanobis distances, and it utilizes so we can evaluate the modified CGAM in the context of a the asymmetries of gait parameters obtained from data rehabilitation therapy. This study aims to understand how recorded during human walking. The gait parameters that the modified CGAM metric can be used to evaluate the gait were used in this analysis represent spatial, temporal, and of individuals with stroke. The study data consists of six kinetic parameters. This form of a consolidated metric will subjects who trained on the device for four weeks. Gait help researchers identify overall gait asymmetry and improve parameters and functional clinical measures were collected rehabilitation techniques to provide a well-rounded gait post throughout the training and used in the modified CGAM training. The CGAM metric successfully served as a measure analysis presented here. for overall symmetry with 11 different gait parameters and 3.1. Subjects. All subjects agreed to participate in this study successfully showed differences among gait with multiple and signed a consent form that was approved by the Western physical asymmetries [14]. The mass at the distal end had a Institutional Review Board. Six subjects (4 males and 2 larger magnitude on overall gait asymmetry compared to females), aged 57–74 years old with unilateral stroke, com- leg length discrepancy. Combined effects are varied based pleted the training, and the length of time since stroke ranged on the cancellation effect between gait parameters [13]. The from 1.2 to 12.5 years. Subject 3 was an outlier and excluded metric was successful in delineating the differences of in some of the analyses. At baseline, his double limb support prosthetic gait and able-bodied gait at three different walking asymmetry was 34 standard deviations above the other velocities [14]. subjects’ mean and timed up and go (TUG) score was 36 Symmetry is calculated using equation (1) where M is standard deviations above the other subjects’ mean. the step length, step time, swing time, double limb support (DLS), and ground reaction forces (GRFs). A value of 0 3.2. Device Used for Gait Training. The device, shown in indicates symmetry. The measures include gait evaluations Figure 2, is designed to change interlimb coordination and conducted before training and after the completion of strengthen the paretic leg of individuals with asymmetric training. walking patterns caused by stroke. The concept of this device is similar to that of a split-belt treadmill [32] but allows the abs M − M paretic nonparetic individual to walk over ground, which is hypothesized to help Symmetry = 100 ∗ , ð1Þ 0:5 ∗ M + M with long-term retention of the altered gait pattern [33]. The paretic nonparetic device is completely passive and uses spiral-like (noncon- sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi stant radius) wheels [34], which redirect the downward force Data ∗ invðÞ Σ ∗ Data ð2Þ Modified CGAM = , generated during walking into a backward force that gener- ∑ inv Σ ðÞ ðÞ ates a consistent motion. By not utilizing actuators and fabri- cating the shoe using rapid manufactured glass-filled nylon, where the version used in this study weighs approximately 900 g. Small unidirectional dampers on the front and back axles (i) Modified CGAM distance: weighted distance from prevent uncontrolled motions. After the shoe stops moving ideal symmetry backward, the user pushes off, and springs attached to the (ii) CGAM distance: Mahalanobis distance from ideal axles reset the position of the wheels for the next step. The symmetry front of the device is able to pivot to more naturally conform to the user’s toe-off. (iii) Data: matrix with n columns (11) and m rows (number of steps) 3.3. Experiment Procedure. Before training, the subject’s gait patterns were evaluated using a ProtoKinetics Zeno Walkway (iv) Σ: covariance of the data (ProtoKinetics, Havertown, PA). They then completed four The modified CGAM [30] works similar to weighted weeks of training three times a week under the guidance of means, but, in this case, the weights are inverse covariances a physical therapist. Each of the twelve sessions included six 4 Applied Bionics and Biomechanics (a) (b) (c) (d) (e) (f ) (g) (h) Figure 2: As the wearer takes a step, the device pushes the foot backward during stance. This exaggeration of the asymmetry results in a more symmetric gait pattern once the shoe is removed. In addition, the shoe works to strengthen the paretic leg by slightly destabilizing the nonparetic leg, which encourages the wearer to use their paretic leg more. A flexible height- and weight-matched platform worn on the opposite foot equalizes the added height and weight of the device. bouts of walking for five minutes on the device with about a and data from all weeks. The ground reaction force has a two-minute break between bouts. The device was attached to stronger correlation for all midtests compared to just the the subject’s nonparetic foot during training. The subject’s pre- and post tests. gait without the device was measured on the ProtoKinetics Table 2 shows the complete list of r values comparing Zeno Walkway before the training began [35]; this data will the gait parameters and modified CGAM to the functional be referred to henceforth as pretest. Gait data was also col- gait measures. Modified CGAM scores show a moderate cor- lected on the walkway prior to the second, third, and fourth relation to TUG and strong correlations with 6MWT and gait week of training sessions; this data will be referred to as velocity. Step time and swing time asymmetries show a midtest. Their gait was tested again within five days after similar pattern of correlation as the modified CGAM does. the completion of the training protocol on the walkway; this TUG shows a moderate correlation to step time, swing time, data will be referred to as post test. Clinical measures and ground reaction force asymmetries, but weak and very included TUG [36], six-minute walk test (6MWT) [37], and weak correlations to step length and double limb support gait velocity. asymmetries, respectively. The 6MWT and gait velocity show moderate correlations to step length asymmetry and strong 3.4. Data Analysis. The modified CGAM scores for all the tri- correlations to step time and swing time asymmetries, but als were calculated using spatial, temporal, and kinetic weak correlations to double limb support and ground reac- parameter asymmetries. The R-squared (r ) was used to tion force asymmetries. assess the correlations between the modified CGAM scores and clinical measures. The correlations between the clinical 5. Discussion measures and individual gait parameters were also analyzed using r . The strength of correlation was evaluated based on Comparing the behavior of the gait parameters helps under- the absolute value of r as reported by Swinscow et al. [38] stand the relationship between the gait asymmetries and also where r =0:4 and above is moderate or strong correlation. evaluates the hypothesis that there exists a balance of asym- metry between gait parameters. For example, most subjects in midtest 1 show a decrease in spatial and temporal 4. Results asymmetry but have increases in ground reaction force The individual gait parameter asymmetries are shown in asymmetry. The reverse is observed in midtest 2 where most Figure 3 for reference. Details related to the results from the subjects have decreased ground reaction force but increased spatial and temporal asymmetry. Not all subjects display clinical trial are presented in another paper [31]. The below results focus on the modified CGAM. the same changes, but this highlights the difficultly of deter- Table 1 shows the correlation values between the pre- and mining if the overall gait improved or not since improving post test data of each gait parameter for all subjects correlated one gait parameter may come partially at the expense of mak- with the corresponding modified CGAM scores. The pre- ing another gait parameter worse. People with hemiparesis and post test performance is important clinically; however due to stroke have different force and motion capabilities it is also important to analyze the correlation for all the on each leg. The paretic leg is weaker and has a more limited midtest data points for the gait parameters, so both time range of motion than the nonparetic leg. Rehabilitation sci- frames are shown. It is interesting to note that step length, ence has not advanced to the point where these problems step time, and swing time show consistently very strong cor- can be fully corrected. Therefore, when we are retraining relation to the modified CGAM while double limb support walking poststroke, we are working with an inherently asym- asymmetry shows a very weak correlation. The correlations metric system. From a biomechanical view, two physically dif- between step length, step time, swing time, and double limb ferent systems (e.g., legs) can only have the same motion if the support remain consistent between the pre-/post comparison forces controlling them or the forces resulting from the Applied Bionics and Biomechanics 5 Step length Step time Double limb support Swing time 40 80 20 40 10 20 0 0 Vertical ground reaction force CGAM 0 0 Subject 1 Subject 5 Subject 6 Subject 2 Subject 4 Figure 3: Gait parameter asymmetry. movement are different. When an individual with an asym- All subjects decreased the modified CGAM score, which metric impairment walks with symmetric step lengths, other indicates that their overall gait improved. This does not mean aspects of gait become asymmetric, such as the forces in the that every gait parameter improved. For example, subject 2 joints [39, 40], the amount of time standing on each leg [21], had slightly worse swing time and vertical ground reaction and other temporal variables [41, 42], all of which can be det- force asymmetries and subject 4 had slightly worse step time rimental to efficiency and long-term viability. and swing time asymmetries during the post test compared Midtest 1 Midtest 1 Midtest 1 Midtest 1 Midtest 1 Midtest 1 Pretest Pretest Pretest Pretest Pretest Pretest Post test Post test Post test Post test Post test Post test Percent asymmetry Percent asymmetry Percent asymmetry CGAM magnitude Percent asymmetry Percent asymmetry Midtest 3 Midtest 3 Midtest 3 Midtest 3 Midtest 3 Midtest 3 Midtest 2 Midtest 2 Midtest 2 Midtest 2 Midtest 2 Midtest 2 6 Applied Bionics and Biomechanics to the underlying parameters, having moderate to strong Table 1: Correlation (r ) between modified CGAM and gait parameters. correlation with the functional measures shows evidence that a measure of overall symmetry which is used as factor Gait parameter Modified CGAM Modified CGAM for gait quality is related to gait function signified by gait (asymmetry) (pre & post) (all midtests) velocity and 6MWT. These findings also offer some evidence Step length 0.93 0.81 to validate the modified CGAM metric. Step time 0.95 0.88 Swing time 0.98 0.89 6. Conclusions Double limb support 0.01 0.01 To summarize, the research suggests that rehabilitating gait Ground reaction force 0.03 0.18 asymmetries should be a holistic approach. Targeting certain Bold implies correlation that is moderate or above. types of asymmetry may not be the correct approach as it may adversely affect other gait parameters that may lead to pervasive long-term effects. The modified CGAM metric showed potential for being used as a quantitative metric for Table 2: Correlation (r ) between clinical measures and gait impairments that cause gait asymmetries. Further, the parameters. research suggests that it is important to consider quantitative metrics such as modified CGAM and subjective metrics such Gait parameter TUG 6MWT Gait velocity as pain and quality of life data to evaluate overall improve- Step length asymmetry 0.14 0.21 0.31 ment of an individual’s gait. The simple asymmetric pertur- Step time asymmetry 0.23 0.53 0.63 bations applied on the gait patterns showed that it is Swing time asymmetry 0.29 0.43 0.57 possible to combat the negative effects of asymmetric impair- Double limb support asymmetry 0.03 0.14 0.10 ment with asymmetry. To tackle these problems, this research has shown that quantitative metrics along with Ground reaction force asymmetry 0.26 0.14 0.13 clinical evaluation offer a good direction in evaluating and Modified CGAM 0.22 0.41 0.51 rehabilitating asymmetric gait patterns. Bold implies correlation that is moderate or above. Data Availability to the pretest. But, the other gait parameters improved such The data used to support the findings of this study are that the end result was an overall better gait pattern. This available from the corresponding author upon request. suggests that there can be a functional balance between all the gait parameters. Although the resulting gait will have some degree of asymmetry in all measures, it will more Conflicts of Interest likely meet the functional walking goals of individuals with K. B. Reed has a licensed patent (US 9,295,302) related to the asymmetric impairments. rehabilitation device used in this work. A management plan The modified CGAM can be calculated using any num- has been implemented and followed to reduce any effects of ber of input gait parameters. Including more should give a this conflict of interest. better indication of the overall gait, but care should be given to including a range of different types of parameters like Acknowledgments forces, spatial, and temporal parameters. Also of note is that the specific score of modified CGAM with one set of Portions of this work have been published in Ramakrishnan’s parameters is not directly comparable to modified CGAM PhD dissertation [29]. Funding for this research has been computed with a different set of parameters. So, modified provided by the Florida High Tech Corridor. This material CGAM can be very helpful for looking at changes within a is based upon work supported by the USA National Science study but may not always provide a comparison between Foundation under Grant Number IIS-1910434. studies if the measured parameters are different. Modified CGAM shows a strong correlation with step References length, step time, and swing time. This was consistent when only the pre- and post test data were considered or when all [1] J. B. Dingwell and B. L. Davis, “A rehabilitation treadmill with test data including pre- and post tests were analyzed. This software for providing real-time gait analysis and visual feed- means that these three parameters have similar behaviors to back,” Journal of Biomechanical Engineering, vol. 118, no. 2, their modified CGAM scores while double limb support pp. 253–255, 1996. and ground reaction force asymmetry have more variation [2] I. Handžić and K. B. Reed, “Perception of gait patterns that in the data. deviate from normal and symmetric biped locomotion,” The modified CGAM scores calculated using the spatial, Frontiers in Psychology, vol. 6, p. 199, 2015. temporal, and kinetic parameters showed behaviors similar [3] D. M. Wrisley, G. F. Marchetti, D. K. Kuharsky, and S. L. to some of the underlying gait parameter asymmetries Whitney, “Reliability, internal consistency, and validity of (see Figure 3) and also some of the functional measures. data obtained with the functional gait assessment,” Physical Although it would be expected to have some correlation Therapy, vol. 82, no. 10, pp. 906–918, 2004. Applied Bionics and Biomechanics 7 measure of paretic leg contribution in hemiparetic walking,” [4] J. McConvey and S. E. Bennett, “Reliability of the dynamic gait index in individuals with multiple sclerosis,” Archives of Phys- Stroke, vol. 37, no. 3, pp. 872–876, 2006. ical Medicine and Rehabilitation, vol. 86, no. 1, pp. 130–133, [21] C. M. Kim and J. J. Eng, “Symmetry in vertical ground reaction force is accompanied by symmetry in temporal but not dis- [5] L. M. Schutte, U. Narayanan, J. L. Stout, P. Selber, J. R. Gage, tance variables of gait in persons with stroke,” Gait & Posture, and M. H. Schwartz, “An index for quantifying deviations vol. 18, no. 1, pp. 23–28, 2003. from normal gait,” Gait & Posture, vol. 11, no. 1, pp. 25–31, [22] D. S. Reisman, H. McLean, and A. J. Bastian, “Split-belt tread- mill training post-stroke: a case study,” Journal of Neurologic [6] M. H. Schwartz and A. Rozumalski, “The gait deviation index: Physical Therapy, vol. 34, no. 4, pp. 202–207, 2010. a new comprehensive index of gait pathology,” Gait & Posture, [23] D. S. Reisman, R. Wityk, K. Silver, and A. J. Bastian, “Split-belt vol. 28, no. 3, pp. 351–357, 2008. treadmill adaptation transfers to overground walking in per- [7] A. Rozumalski and M. H. Schwartz, “The GDI-kinetic: a new sons poststroke,” Neurorehabilitation and Neural Repair, index for quantifying kinetic deviations from normal gait,” vol. 23, no. 7, pp. 735–744, 2009. Gait & Posture, vol. 33, no. 4, pp. 730–732, 2011. [24] J. Mehrholz, S. Thomas, and B. Elsner, “Treadmill training and [8] V. L. Chester, M. Tingley, and E. N. Biden, “An extended index body weight support for walking after stroke,” Cochrane Data- to quantify normality of gait in children,” Gait & Posture, base of Systematic Reviews, vol. 2017, no. 8, article CD002840, vol. 25, no. 4, pp. 549–554, 2007. [9] S. Hoerzer, P. A. Federolf, C. Maurer, J. Baltich, and B. M. [25] P.-C. Kao, S. Srivastava, S. K. Agrawal, and J. P. Scholz, “Effect Nigg, “Footwear decreases gait asymmetry during running,” of robotic performance-based error-augmentation versus PLoS One, vol. 10, no. 10, article e0138631, 2015. error-reduction training on the gait of healthy individuals,” Gait & Posture, vol. 37, no. 1, pp. 113–120, 2013. [10] M. Schlafly, Y. Yilmaz, and K. B. Reed, “Feature selection in gait classification of leg length and distal mass,” Informatics [26] J. J. Daly, J. Zimbelman, K. L. Roenigk et al., “Recovery of in Medicine Unlocked, vol. 15, p. 100163, 2019. coordinated gait: randomized controlled stroke trial of functional electrical stimulation (fes) versus no fes, with [11] A. M. S. Muniz and J. Nadal, “Application of principal compo- weight-supported treadmill and over-ground training,” nent analysis in vertical ground reaction force to discriminate Neurorehabilitation and Neural Repair, vol. 25, no. 7, normal and abnormal gait,” Gait & Posture, vol. 29, no. 1, pp. 588–596, 2011. pp. 31–35, 2009. [27] R.-Y. Wang, F.-Y. Wang, S.-F. Huang, and Y.-R. Yang, “High- [12] T. Ramakrishnan, M. Schlafly, and K. B. Reed, “Effect of asym- frequency repetitive transcranial magnetic stimulation metric knee height on gait asymmetry for unilateral transfe- enhanced treadmill training effects on gait performance in moral amputees,” International Journal of Current Advanced individuals with chronic stroke: a double-blinded randomized Research, vol. 6, no. 10, p. 6896, 2017. controlled pilot trial,” Gait & Posture, vol. 68, pp. 382–387, [13] H. Muratagic, T. Ramakrishnan, and K. B. Reed, “Combined effects of leg length discrepancy and the addition of distal mass on gait asymmetry,” Gait & Posture, [28] P. Cheng and P. Lai, “Comparison of exoskeleton robots and vol. 58, article end-effector robots on training methods and gait biomechan- S0966636217309086, pp. 487–492, 2017. ics,” in International Conference on Intelligent Robotics and [14] T. Ramakrishnan, C.-A. Lahiff, and K. B. Reed, “Comparing Applications, pp. 258–266, Springer, 2013. gait with multiple physical asymmetries using consolidated [29] T. Ramakrishnan, Rehabilitating asymmetric gait using asym- metrics,” Frontiers in Neurorobotics, vol. 12, p. 2, 2018. metry, [Ph.D. thesis], University of South Florida library, [15] M. Brandstater, H. de Bruin, C. Gowland, and B. Clark, Tampa, FL, USA, 2017. “Hemiplegic gait: analysis of temporal variables,” Archives of [30] T. Ramakrishnan, H. Muratagic, and K. B. Reed, “Combined Physical Medicine and Rehabilitation, vol. 64, no. 12, pp. 583–587, 1983. gait asymmetry metric,” in 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology [16] M. Kelly-Hayes, A. Beiser, C. S. Kase, A. Scaramucci, R. B. Society (EMBC), Orlando, FL, USA, August 2016. D’Agostino, and P. A. Wolf, “The influence of gender and [31] S. H. Kim, D. E. Huizenga, I. Handzic et al., “Relearning func- age on disability following ischemic stroke: the Framingham study,” Journal of Stroke and Cerebrovascular Diseases, tional and symmetric walking after stroke using a wearable device: a feasibility study,” Journal of Neuroengineering and vol. 12, no. 3, pp. 119–126, 2003. Rehabilitation, vol. 16, no. 1, pp. 1–8, 2019. [17] J. C. Wall and G. I. Turnbull, “Gait asymmetries in residual [32] D. S. Reisman and A. J. Bastian, “Split-belt treadmill hemiplegia,” Archives of Physical Medicine and Rehabilitation, vol. 67, no. 8, pp. 550–553, 1986. adaptation and GAIT SYMMETRY post-stroke,” Journal of Neurologic Physical Therapy, vol. 29, no. 4, p. 196, 2005. [18] E. B. Titianova and I. M. Tarkka, “Asymmetry in walking per- formance and postural sway in patients with chronic unilateral [33] I. Handzić, E. Vasudevan, and K. B. Reed, “Developing a cerebral infarction,” Journal of Rehabilitation Research and gait enhancing mobile shoe to alter over-ground walking Development, vol. 32, pp. 236–244, 1995. coordination,” in 2012 IEEE International Conference on Robotics and Automation, pp. 4129–4142, Saint Paul, MN, [19] C. K. Balasubramanian, M. G. Bowden, R. R. Neptune, and USA, May 2012. S. A. Kautz, “Relationship between step length asymmetry and walking performance in subjects with chronic hemipar- [34] I. Handzic and K. B. Reed, “Kinetic shapes: analysis, verifica- esis,” Archives of Physical Medicine and Rehabilitation, tion, and applications,” Journal of Mechanical Design, vol. 88, no. 1, pp. 43–49, 2007. vol. 136, no. 6, article 061005, pp. 0610051–0610058, 2014. [20] M. G. Bowden, C. K. Balasubramanian, R. R. Neptune, and [35] R. C. Lynall, L. A. Zukowski, P. Plummer, and J. P. Mihalik, S. A. Kautz, “Anterior-posterior ground reaction forces as a “Reliability and validity of the protokinetics movement 8 Applied Bionics and Biomechanics analysis software in measuring center of pressure during walk- ing,” Gait & Posture, vol. 52, article S0966636216307111, pp. 308–311, 2017. [36] S. S. Ng and C. W. Hui-Chan, “The timed up & go test: its reliability and association with lower-limb impairments and locomotor capacities in people with chronic stroke,” Archives of Physical Medicine and Rehabilitation, vol. 86, no. 8, pp. 1641–1647, 2005. [37] P. S. Pohl, P. W. Duncan, S. Perera et al., “Influence of stroke- related impairments on performance in 6-minute walk test,” Journal of Rehabilitation Research and Development, vol. 39, no. 4, pp. 439–444, 2002. [38] T. D. V. Swinscow and M. J. Campbell, Statistics at square one, BMJ, London, 2002. [39] F. P. Carpes, C. B. Mota, and I. E. Faria, “On the bilateral asym- metry during running and cycling - A review considering leg preference,” Physical Therapy in Sport, vol. 11, no. 4, pp. 136–142, 2010. [40] I. Handzic, H. Muratagic, and K. Reed, “Passive kinematic syn- chronization of dissimilar and uncoupled rotating systems,” Nonlinear Dynamics and Systems Theory, vol. 15, pp. 383– 399, 2015. [41] H. Sadeghi, P. Allard, F. Prince, and H. Labelle, “Symmetry and limb dominance in able-bodied gait: a review,” Gait & Posture, vol. 12, no. 1, pp. 34–45, 2000. [42] M. J. Highsmith, J. T. Kahle, S. L. Carey et al., “Kinetic asym- metry in transfemoral amputees while performing sit to stand and stand to sit movements,” Gait & Posture, vol. 34, no. 1, pp. 86–91, 2011. International Journal of Advances in Rotating Machinery Multimedia Journal of The Scientific Journal of Engineering World Journal Sensors Hindawi Hindawi Publishing Corporation Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 http://www www.hindawi.com .hindawi.com V Volume 2018 olume 2013 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Journal of Control Science and Engineering Advances in Civil Engineering Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Submit your manuscripts at www.hindawi.com Journal of Journal of Electrical and Computer Robotics Engineering Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 VLSI Design Advances in OptoElectronics International Journal of Modelling & Aerospace International Journal of Simulation Navigation and in Engineering Engineering Observation Hindawi Hindawi Hindawi Hindawi Volume 2018 Volume 2018 Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com www.hindawi.com www.hindawi.com Volume 2018 International Journal of Active and Passive International Journal of Antennas and Advances in Chemical Engineering Propagation Electronic Components Shock and Vibration Acoustics and Vibration Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Applied Bionics and Biomechanics Hindawi Publishing Corporation

Human Gait Analysis Metric for Gait Retraining

Loading next page...
 
/lp/hindawi-publishing-corporation/human-gait-analysis-metric-for-gait-retraining-SRA5R0mwbL
Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2019 Tyagi Ramakrishnan et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
1176-2322
eISSN
1754-2103
DOI
10.1155/2019/1286864
Publisher site
See Article on Publisher Site

Abstract

Hindawi Applied Bionics and Biomechanics Volume 2019, Article ID 1286864, 8 pages https://doi.org/10.1155/2019/1286864 Research Article Tyagi Ramakrishnan, Seok Hun Kim, and Kyle B. Reed University of South Florida, USA Correspondence should be addressed to Kyle B. Reed; kylereed@usf.edu Received 19 April 2019; Revised 25 July 2019; Accepted 10 September 2019; Published 11 November 2019 Guest Editor: Michelle Johnson Copyright © 2019 Tyagi Ramakrishnan et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The combined gait asymmetry metric (CGAM) provides a method to synthesize human gait motion. The metric is weighted to balance each parameter’seffect by normalizing the data so all parameters are more equally weighted. It is designed to combine spatial, temporal, kinematic, and kinetic gait parameter asymmetries. It can also combine subsets of the different gait parameters to provide a more thorough analysis. The single number quantifying gait could assist robotic rehabilitation methods to optimize the resulting gait patterns. CGAM will help define quantitative thresholds for achievable balanced overall gait asymmetry. The study presented here compares the combined gait parameters with clinical measures such as timed up and go (TUG), six-minute walk test (6MWT), and gait velocity. The comparisons are made on gait data collected on individuals with stroke before and after twelve sessions of rehabilitation. Step length, step time, and swing time showed a strong correlation to CGAM, but the double limb support asymmetry has nearly no correlation with CGAM and ground reaction force asymmetry has a weak correlation. The CGAM scores were moderately correlated with TUG and strongly correlated to 6MWT and gait velocity. 1. Introduction changes to the individual’s gait pattern from baseline to after the clinical intervention. Figure 1 shows an example of how a combined metric Researchers traditionally analyze a small set of gait parame- would be useful in analyzing an asymmetric gait pattern. ters in order to evaluate the outcomes of their techniques. This often leads to an overreliance on a few parameters and Many existing rehabilitation therapies can change different sets of gait parameters, but some make one parameter worse a focus on improving one gait parameter. Few studies in the while correcting others. Even in unimpaired walking, perfect gait literature aim to correct many gait parameters at the symmetry is not expected [1], so there is space for some same time. This traditional narrow approach lacks broader parameters to be asymmetric while the overall gait is within understanding of the interaction between various gait param- a reasonable bound. The CGAM distance (shown in orange eters and limits potential approaches that can lead to whole- in Figure 1) generates a single representation of the measured some rehabilitation techniques. In this research study, we gait parameters that generally scales with the global deviation examine our combined gait asymmetry metric (CGAM) to give a representation of the overall gait pattern. We use from symmetry. The deviation of each measure is scaled based on the variance within that measure, so measures that stroke for examining this combined metric because it affects generally have larger magnitudes of asymmetry (e.g., forces) several different aspects of an individual’s gait, and many of will be scaled so that each gait parameter has a similar influ- these aspects are asymmetric. Although we focus on mea- ence on the overall metric. If a therapy reduces the CGAM sures of asymmetry, this combined method is not limited by the type or number of parameters evaluated. Our hypoth- distance, the overall gait has improved even though some of the individual parameters might have gotten worse. Without esis is that the outcomes of the combined metric will partially a combined metric, it is difficult to determine whether the correlate to functional clinical outcome measures. We also gait is improving when looking at individual gait parameters. use this combined metric to determine if there have been 2 Applied Bionics and Biomechanics analysis like principle component analysis (PCA) and singu- Rehabilitated Asymmetric gait pattern 1 lar variable decomposition (SVD) to reduce dimensionality gait pattern to make the data computation easier [11]. The processed data is then classified using the Euclidean or similar distances [11]. These distances become the scores which form the cen- tral part of the gait metric. Another study by Hoerzer et al. [9] proposed the comprehensive asymmetry index (CAI) which Rehabilitated combined gait asymmetry using PCA and Euclidean dis- Rehabilitated gait pattern 3 tances. CAI was effective in identifying that running with gait pattern 2 shoes reduces gait asymmetry compared to barefoot running. Gait parameter 2, A prior study used a combination of Mahalanobis distances e.g., step time asymmetry with data reduction techniques on a preprocessed dataset to analyze kinematic and kinetic gait parameters [8]. They Perfect symmetry developed several metrics to classify the data and showed that they can successfully classify the abnormal data from a standard normal dataset. The precursor to CGAM used Typical able-body gait (slightly asymmetric) a symmetry index processed using PCA measured using Mahalanobis distances. Without the restrictions of dimen- sionality reduction, CGAM served as a versatile gait asymme- Figure 1: Representation of the multidimensional gait parameter space. The orange lines represent the distance each gait is from a try metric [12–14]. symmetric gait (CGAM distance), which helps determine how far away a gait is from ideal. CGAM can also aid in ascertaining 1.3. Effects of Stroke on Gait and Rehabilitation. The analysis whether the overall gait pattern is improving (even if some of the in this paper uses an existing dataset from an experimental parameters are getting worse). CGAM can incorporate more stroke therapy to examine the effects of combining and dimensions than the three shown, but that is hard to visualize. jointly assessing gait as opposed to individually assessing a single parameter. We focus on individuals with stroke 1.1. Gait Measurements. Gait data is typically collected using because they inherently have different capabilities on each motion capture, force plates, and/or wearable sensors. Many side and are asymmetric; as such, it is unlikely that they can ever regain complete symmetry in all parameters. However, variables portray various facets of human gait. There are spa- tial parameters such as step length defined by the distance it may be possible to achieve a balanced gait where some covered from the heel strike of one foot to the heel strike of parameters are slightly asymmetric, but none of them are the opposite foot. There are temporal parameters such as step excessively large. Our proposed joint metric helps to balance time defined as the time taken between opposite heel strikes. all of the parameters. We examine before and after the therapy to help understand what changes have occurred. Then, there is swing time, which is the time taken from toe- off to heel strike of the same foot. Double limb support is Gait after stroke becomes asymmetric (or hemiparetic) as the time spent when both legs are on the ground. The termi- a consequence of altered neuromuscular signals affecting leg nal double limb support is used for this research study. There motor areas, typically hyperextension at the knee and are kinematic parameters associated with joint angles of the reduced flexion at the hip, knee, and ankle [15–17]. Hemi- paretic gait is characterized by significant asymmetry in ankle, knee, and hip joints. Hip joints in the case of individ- uals with stroke and amputees also show abduction and temporal (e.g., time spent in double limb support) and adduction. The kinetic parameters include vertical ground spatial (e.g., step length) measures of interlimb coordination reaction forces, propulsive or push-off forces during toe-off, [15, 18, 19]. Propulsive force of the paretic limb is reduced braking forces during initial contact or heel strike, and ankle, compared to the nonparetic limb, as are work and power of the paretic plantar flexors [19, 20]. The significant decrease knee, and hip joint moments. Further, some of these param- eters are more easily identified by sight alone (e.g., step in propulsive force results in smaller overall step lengths, length, cadence, and gait velocity) while others are nearly which in turn affects the patient’s gait velocity. Finally, verti- impossible to quantify without a sensor (e.g., forces and joint cal ground reaction forces (GRFs) are decreased on the moments) [2]. paretic limb relative to the nonparetic limb [21], reflecting diminished weight bearing and balancing capabilities by the 1.2. Gait Metrics. Several gait metrics combining multiple paretic limb. gait parameters have been used clinically to evaluate different Some of the rehabilitation techniques used to restore gait gait impairments. These metrics can also be used to classify impaired by stroke involve some form of asymmetric pertur- gait based on different types of information. There are two bations that try to restore the symmetry between the paretic and nonparetic sides [22]. Split-belt treadmills are one types: qualitative [3, 4] and quantitative [5–7] metrics. Many metrics rely on either kinetic or kinematic data to categorize method to apply this rehabilitation technique. The split-belt different gait motions and behaviors. Some metrics have the treadmill has two treads that can move at different velocities, ability to jointly analyze kinetic and kinematic parameters which are used to exaggerate the asymmetry of the individ- [8, 9]. Machine learning has been used to classify and differ- ual. When the tread speeds are made the same after training, the subject typically has some after-effects that are more entiate gait patterns [10]. Most gait metrics use statistical CGAM metric Gait parameter 3, e.g., push-off force Gait parameter 1, e.g., step length Applied Bionics and Biomechanics 3 that are multiplied across the dataset in the numerator. To symmetric than when they started [23]. The after-effects are usually improved spatial and temporal symmetry. Unfortu- balance the influence of the inverse of covariance, it is divided nately, these after-effects only partially transfer to walking by the sum of the inverse covariance matrix, equation (2). on the ground. There are other rehabilitation techniques This change to the formulation makes the modified CGAM such as body-weight support [24], robotic [25], functional represent the scores closer to the percent asymmetry while electrical stimulation [26], transcranial magnetic stimulation still serving as a combined measure of all the gait parameter [27], and full-body gait exoskeletons [28]. Each of the tech- asymmetries. niques have their merits and train the individual in a special- ized manner, which means a combination of these methods 3. Methods may provide additional benefits to the person. The analysis performed in this paper used data collected as part of a separate clinical study. The novel shoe tested was 2. CGAM Derivation designed to improve the overall gait symmetry and gait func- The metric presented here has the potential to help categorize tion of an individual poststroke. The efficacy of the device is and differentiate between multiple asymmetric gaits [29]. discussed in another paper [31]. That study data is used here CGAM is based on Mahalanobis distances, and it utilizes so we can evaluate the modified CGAM in the context of a the asymmetries of gait parameters obtained from data rehabilitation therapy. This study aims to understand how recorded during human walking. The gait parameters that the modified CGAM metric can be used to evaluate the gait were used in this analysis represent spatial, temporal, and of individuals with stroke. The study data consists of six kinetic parameters. This form of a consolidated metric will subjects who trained on the device for four weeks. Gait help researchers identify overall gait asymmetry and improve parameters and functional clinical measures were collected rehabilitation techniques to provide a well-rounded gait post throughout the training and used in the modified CGAM training. The CGAM metric successfully served as a measure analysis presented here. for overall symmetry with 11 different gait parameters and 3.1. Subjects. All subjects agreed to participate in this study successfully showed differences among gait with multiple and signed a consent form that was approved by the Western physical asymmetries [14]. The mass at the distal end had a Institutional Review Board. Six subjects (4 males and 2 larger magnitude on overall gait asymmetry compared to females), aged 57–74 years old with unilateral stroke, com- leg length discrepancy. Combined effects are varied based pleted the training, and the length of time since stroke ranged on the cancellation effect between gait parameters [13]. The from 1.2 to 12.5 years. Subject 3 was an outlier and excluded metric was successful in delineating the differences of in some of the analyses. At baseline, his double limb support prosthetic gait and able-bodied gait at three different walking asymmetry was 34 standard deviations above the other velocities [14]. subjects’ mean and timed up and go (TUG) score was 36 Symmetry is calculated using equation (1) where M is standard deviations above the other subjects’ mean. the step length, step time, swing time, double limb support (DLS), and ground reaction forces (GRFs). A value of 0 3.2. Device Used for Gait Training. The device, shown in indicates symmetry. The measures include gait evaluations Figure 2, is designed to change interlimb coordination and conducted before training and after the completion of strengthen the paretic leg of individuals with asymmetric training. walking patterns caused by stroke. The concept of this device is similar to that of a split-belt treadmill [32] but allows the abs M − M paretic nonparetic individual to walk over ground, which is hypothesized to help Symmetry = 100 ∗ , ð1Þ 0:5 ∗ M + M with long-term retention of the altered gait pattern [33]. The paretic nonparetic device is completely passive and uses spiral-like (noncon- sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi stant radius) wheels [34], which redirect the downward force Data ∗ invðÞ Σ ∗ Data ð2Þ Modified CGAM = , generated during walking into a backward force that gener- ∑ inv Σ ðÞ ðÞ ates a consistent motion. By not utilizing actuators and fabri- cating the shoe using rapid manufactured glass-filled nylon, where the version used in this study weighs approximately 900 g. Small unidirectional dampers on the front and back axles (i) Modified CGAM distance: weighted distance from prevent uncontrolled motions. After the shoe stops moving ideal symmetry backward, the user pushes off, and springs attached to the (ii) CGAM distance: Mahalanobis distance from ideal axles reset the position of the wheels for the next step. The symmetry front of the device is able to pivot to more naturally conform to the user’s toe-off. (iii) Data: matrix with n columns (11) and m rows (number of steps) 3.3. Experiment Procedure. Before training, the subject’s gait patterns were evaluated using a ProtoKinetics Zeno Walkway (iv) Σ: covariance of the data (ProtoKinetics, Havertown, PA). They then completed four The modified CGAM [30] works similar to weighted weeks of training three times a week under the guidance of means, but, in this case, the weights are inverse covariances a physical therapist. Each of the twelve sessions included six 4 Applied Bionics and Biomechanics (a) (b) (c) (d) (e) (f ) (g) (h) Figure 2: As the wearer takes a step, the device pushes the foot backward during stance. This exaggeration of the asymmetry results in a more symmetric gait pattern once the shoe is removed. In addition, the shoe works to strengthen the paretic leg by slightly destabilizing the nonparetic leg, which encourages the wearer to use their paretic leg more. A flexible height- and weight-matched platform worn on the opposite foot equalizes the added height and weight of the device. bouts of walking for five minutes on the device with about a and data from all weeks. The ground reaction force has a two-minute break between bouts. The device was attached to stronger correlation for all midtests compared to just the the subject’s nonparetic foot during training. The subject’s pre- and post tests. gait without the device was measured on the ProtoKinetics Table 2 shows the complete list of r values comparing Zeno Walkway before the training began [35]; this data will the gait parameters and modified CGAM to the functional be referred to henceforth as pretest. Gait data was also col- gait measures. Modified CGAM scores show a moderate cor- lected on the walkway prior to the second, third, and fourth relation to TUG and strong correlations with 6MWT and gait week of training sessions; this data will be referred to as velocity. Step time and swing time asymmetries show a midtest. Their gait was tested again within five days after similar pattern of correlation as the modified CGAM does. the completion of the training protocol on the walkway; this TUG shows a moderate correlation to step time, swing time, data will be referred to as post test. Clinical measures and ground reaction force asymmetries, but weak and very included TUG [36], six-minute walk test (6MWT) [37], and weak correlations to step length and double limb support gait velocity. asymmetries, respectively. The 6MWT and gait velocity show moderate correlations to step length asymmetry and strong 3.4. Data Analysis. The modified CGAM scores for all the tri- correlations to step time and swing time asymmetries, but als were calculated using spatial, temporal, and kinetic weak correlations to double limb support and ground reac- parameter asymmetries. The R-squared (r ) was used to tion force asymmetries. assess the correlations between the modified CGAM scores and clinical measures. The correlations between the clinical 5. Discussion measures and individual gait parameters were also analyzed using r . The strength of correlation was evaluated based on Comparing the behavior of the gait parameters helps under- the absolute value of r as reported by Swinscow et al. [38] stand the relationship between the gait asymmetries and also where r =0:4 and above is moderate or strong correlation. evaluates the hypothesis that there exists a balance of asym- metry between gait parameters. For example, most subjects in midtest 1 show a decrease in spatial and temporal 4. Results asymmetry but have increases in ground reaction force The individual gait parameter asymmetries are shown in asymmetry. The reverse is observed in midtest 2 where most Figure 3 for reference. Details related to the results from the subjects have decreased ground reaction force but increased spatial and temporal asymmetry. Not all subjects display clinical trial are presented in another paper [31]. The below results focus on the modified CGAM. the same changes, but this highlights the difficultly of deter- Table 1 shows the correlation values between the pre- and mining if the overall gait improved or not since improving post test data of each gait parameter for all subjects correlated one gait parameter may come partially at the expense of mak- with the corresponding modified CGAM scores. The pre- ing another gait parameter worse. People with hemiparesis and post test performance is important clinically; however due to stroke have different force and motion capabilities it is also important to analyze the correlation for all the on each leg. The paretic leg is weaker and has a more limited midtest data points for the gait parameters, so both time range of motion than the nonparetic leg. Rehabilitation sci- frames are shown. It is interesting to note that step length, ence has not advanced to the point where these problems step time, and swing time show consistently very strong cor- can be fully corrected. Therefore, when we are retraining relation to the modified CGAM while double limb support walking poststroke, we are working with an inherently asym- asymmetry shows a very weak correlation. The correlations metric system. From a biomechanical view, two physically dif- between step length, step time, swing time, and double limb ferent systems (e.g., legs) can only have the same motion if the support remain consistent between the pre-/post comparison forces controlling them or the forces resulting from the Applied Bionics and Biomechanics 5 Step length Step time Double limb support Swing time 40 80 20 40 10 20 0 0 Vertical ground reaction force CGAM 0 0 Subject 1 Subject 5 Subject 6 Subject 2 Subject 4 Figure 3: Gait parameter asymmetry. movement are different. When an individual with an asym- All subjects decreased the modified CGAM score, which metric impairment walks with symmetric step lengths, other indicates that their overall gait improved. This does not mean aspects of gait become asymmetric, such as the forces in the that every gait parameter improved. For example, subject 2 joints [39, 40], the amount of time standing on each leg [21], had slightly worse swing time and vertical ground reaction and other temporal variables [41, 42], all of which can be det- force asymmetries and subject 4 had slightly worse step time rimental to efficiency and long-term viability. and swing time asymmetries during the post test compared Midtest 1 Midtest 1 Midtest 1 Midtest 1 Midtest 1 Midtest 1 Pretest Pretest Pretest Pretest Pretest Pretest Post test Post test Post test Post test Post test Post test Percent asymmetry Percent asymmetry Percent asymmetry CGAM magnitude Percent asymmetry Percent asymmetry Midtest 3 Midtest 3 Midtest 3 Midtest 3 Midtest 3 Midtest 3 Midtest 2 Midtest 2 Midtest 2 Midtest 2 Midtest 2 Midtest 2 6 Applied Bionics and Biomechanics to the underlying parameters, having moderate to strong Table 1: Correlation (r ) between modified CGAM and gait parameters. correlation with the functional measures shows evidence that a measure of overall symmetry which is used as factor Gait parameter Modified CGAM Modified CGAM for gait quality is related to gait function signified by gait (asymmetry) (pre & post) (all midtests) velocity and 6MWT. These findings also offer some evidence Step length 0.93 0.81 to validate the modified CGAM metric. Step time 0.95 0.88 Swing time 0.98 0.89 6. Conclusions Double limb support 0.01 0.01 To summarize, the research suggests that rehabilitating gait Ground reaction force 0.03 0.18 asymmetries should be a holistic approach. Targeting certain Bold implies correlation that is moderate or above. types of asymmetry may not be the correct approach as it may adversely affect other gait parameters that may lead to pervasive long-term effects. The modified CGAM metric showed potential for being used as a quantitative metric for Table 2: Correlation (r ) between clinical measures and gait impairments that cause gait asymmetries. Further, the parameters. research suggests that it is important to consider quantitative metrics such as modified CGAM and subjective metrics such Gait parameter TUG 6MWT Gait velocity as pain and quality of life data to evaluate overall improve- Step length asymmetry 0.14 0.21 0.31 ment of an individual’s gait. The simple asymmetric pertur- Step time asymmetry 0.23 0.53 0.63 bations applied on the gait patterns showed that it is Swing time asymmetry 0.29 0.43 0.57 possible to combat the negative effects of asymmetric impair- Double limb support asymmetry 0.03 0.14 0.10 ment with asymmetry. To tackle these problems, this research has shown that quantitative metrics along with Ground reaction force asymmetry 0.26 0.14 0.13 clinical evaluation offer a good direction in evaluating and Modified CGAM 0.22 0.41 0.51 rehabilitating asymmetric gait patterns. Bold implies correlation that is moderate or above. Data Availability to the pretest. But, the other gait parameters improved such The data used to support the findings of this study are that the end result was an overall better gait pattern. This available from the corresponding author upon request. suggests that there can be a functional balance between all the gait parameters. Although the resulting gait will have some degree of asymmetry in all measures, it will more Conflicts of Interest likely meet the functional walking goals of individuals with K. B. Reed has a licensed patent (US 9,295,302) related to the asymmetric impairments. rehabilitation device used in this work. A management plan The modified CGAM can be calculated using any num- has been implemented and followed to reduce any effects of ber of input gait parameters. Including more should give a this conflict of interest. better indication of the overall gait, but care should be given to including a range of different types of parameters like Acknowledgments forces, spatial, and temporal parameters. Also of note is that the specific score of modified CGAM with one set of Portions of this work have been published in Ramakrishnan’s parameters is not directly comparable to modified CGAM PhD dissertation [29]. Funding for this research has been computed with a different set of parameters. So, modified provided by the Florida High Tech Corridor. This material CGAM can be very helpful for looking at changes within a is based upon work supported by the USA National Science study but may not always provide a comparison between Foundation under Grant Number IIS-1910434. studies if the measured parameters are different. Modified CGAM shows a strong correlation with step References length, step time, and swing time. This was consistent when only the pre- and post test data were considered or when all [1] J. B. Dingwell and B. L. Davis, “A rehabilitation treadmill with test data including pre- and post tests were analyzed. This software for providing real-time gait analysis and visual feed- means that these three parameters have similar behaviors to back,” Journal of Biomechanical Engineering, vol. 118, no. 2, their modified CGAM scores while double limb support pp. 253–255, 1996. and ground reaction force asymmetry have more variation [2] I. Handžić and K. B. Reed, “Perception of gait patterns that in the data. deviate from normal and symmetric biped locomotion,” The modified CGAM scores calculated using the spatial, Frontiers in Psychology, vol. 6, p. 199, 2015. temporal, and kinetic parameters showed behaviors similar [3] D. M. Wrisley, G. F. Marchetti, D. K. Kuharsky, and S. L. to some of the underlying gait parameter asymmetries Whitney, “Reliability, internal consistency, and validity of (see Figure 3) and also some of the functional measures. data obtained with the functional gait assessment,” Physical Although it would be expected to have some correlation Therapy, vol. 82, no. 10, pp. 906–918, 2004. Applied Bionics and Biomechanics 7 measure of paretic leg contribution in hemiparetic walking,” [4] J. McConvey and S. E. Bennett, “Reliability of the dynamic gait index in individuals with multiple sclerosis,” Archives of Phys- Stroke, vol. 37, no. 3, pp. 872–876, 2006. ical Medicine and Rehabilitation, vol. 86, no. 1, pp. 130–133, [21] C. M. Kim and J. J. Eng, “Symmetry in vertical ground reaction force is accompanied by symmetry in temporal but not dis- [5] L. M. Schutte, U. Narayanan, J. L. Stout, P. Selber, J. R. Gage, tance variables of gait in persons with stroke,” Gait & Posture, and M. H. Schwartz, “An index for quantifying deviations vol. 18, no. 1, pp. 23–28, 2003. from normal gait,” Gait & Posture, vol. 11, no. 1, pp. 25–31, [22] D. S. Reisman, H. McLean, and A. J. Bastian, “Split-belt tread- mill training post-stroke: a case study,” Journal of Neurologic [6] M. H. Schwartz and A. Rozumalski, “The gait deviation index: Physical Therapy, vol. 34, no. 4, pp. 202–207, 2010. a new comprehensive index of gait pathology,” Gait & Posture, [23] D. S. Reisman, R. Wityk, K. Silver, and A. J. Bastian, “Split-belt vol. 28, no. 3, pp. 351–357, 2008. treadmill adaptation transfers to overground walking in per- [7] A. Rozumalski and M. H. Schwartz, “The GDI-kinetic: a new sons poststroke,” Neurorehabilitation and Neural Repair, index for quantifying kinetic deviations from normal gait,” vol. 23, no. 7, pp. 735–744, 2009. Gait & Posture, vol. 33, no. 4, pp. 730–732, 2011. [24] J. Mehrholz, S. Thomas, and B. Elsner, “Treadmill training and [8] V. L. Chester, M. Tingley, and E. N. Biden, “An extended index body weight support for walking after stroke,” Cochrane Data- to quantify normality of gait in children,” Gait & Posture, base of Systematic Reviews, vol. 2017, no. 8, article CD002840, vol. 25, no. 4, pp. 549–554, 2007. [9] S. Hoerzer, P. A. Federolf, C. Maurer, J. Baltich, and B. M. [25] P.-C. Kao, S. Srivastava, S. K. Agrawal, and J. P. Scholz, “Effect Nigg, “Footwear decreases gait asymmetry during running,” of robotic performance-based error-augmentation versus PLoS One, vol. 10, no. 10, article e0138631, 2015. error-reduction training on the gait of healthy individuals,” Gait & Posture, vol. 37, no. 1, pp. 113–120, 2013. [10] M. Schlafly, Y. Yilmaz, and K. B. Reed, “Feature selection in gait classification of leg length and distal mass,” Informatics [26] J. J. Daly, J. Zimbelman, K. L. Roenigk et al., “Recovery of in Medicine Unlocked, vol. 15, p. 100163, 2019. coordinated gait: randomized controlled stroke trial of functional electrical stimulation (fes) versus no fes, with [11] A. M. S. Muniz and J. Nadal, “Application of principal compo- weight-supported treadmill and over-ground training,” nent analysis in vertical ground reaction force to discriminate Neurorehabilitation and Neural Repair, vol. 25, no. 7, normal and abnormal gait,” Gait & Posture, vol. 29, no. 1, pp. 588–596, 2011. pp. 31–35, 2009. [27] R.-Y. Wang, F.-Y. Wang, S.-F. Huang, and Y.-R. Yang, “High- [12] T. Ramakrishnan, M. Schlafly, and K. B. Reed, “Effect of asym- frequency repetitive transcranial magnetic stimulation metric knee height on gait asymmetry for unilateral transfe- enhanced treadmill training effects on gait performance in moral amputees,” International Journal of Current Advanced individuals with chronic stroke: a double-blinded randomized Research, vol. 6, no. 10, p. 6896, 2017. controlled pilot trial,” Gait & Posture, vol. 68, pp. 382–387, [13] H. Muratagic, T. Ramakrishnan, and K. B. Reed, “Combined effects of leg length discrepancy and the addition of distal mass on gait asymmetry,” Gait & Posture, [28] P. Cheng and P. Lai, “Comparison of exoskeleton robots and vol. 58, article end-effector robots on training methods and gait biomechan- S0966636217309086, pp. 487–492, 2017. ics,” in International Conference on Intelligent Robotics and [14] T. Ramakrishnan, C.-A. Lahiff, and K. B. Reed, “Comparing Applications, pp. 258–266, Springer, 2013. gait with multiple physical asymmetries using consolidated [29] T. Ramakrishnan, Rehabilitating asymmetric gait using asym- metrics,” Frontiers in Neurorobotics, vol. 12, p. 2, 2018. metry, [Ph.D. thesis], University of South Florida library, [15] M. Brandstater, H. de Bruin, C. Gowland, and B. Clark, Tampa, FL, USA, 2017. “Hemiplegic gait: analysis of temporal variables,” Archives of [30] T. Ramakrishnan, H. Muratagic, and K. B. Reed, “Combined Physical Medicine and Rehabilitation, vol. 64, no. 12, pp. 583–587, 1983. gait asymmetry metric,” in 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology [16] M. Kelly-Hayes, A. Beiser, C. S. Kase, A. Scaramucci, R. B. Society (EMBC), Orlando, FL, USA, August 2016. D’Agostino, and P. A. Wolf, “The influence of gender and [31] S. H. Kim, D. E. Huizenga, I. Handzic et al., “Relearning func- age on disability following ischemic stroke: the Framingham study,” Journal of Stroke and Cerebrovascular Diseases, tional and symmetric walking after stroke using a wearable device: a feasibility study,” Journal of Neuroengineering and vol. 12, no. 3, pp. 119–126, 2003. Rehabilitation, vol. 16, no. 1, pp. 1–8, 2019. [17] J. C. Wall and G. I. Turnbull, “Gait asymmetries in residual [32] D. S. Reisman and A. J. Bastian, “Split-belt treadmill hemiplegia,” Archives of Physical Medicine and Rehabilitation, vol. 67, no. 8, pp. 550–553, 1986. adaptation and GAIT SYMMETRY post-stroke,” Journal of Neurologic Physical Therapy, vol. 29, no. 4, p. 196, 2005. [18] E. B. Titianova and I. M. Tarkka, “Asymmetry in walking per- formance and postural sway in patients with chronic unilateral [33] I. Handzić, E. Vasudevan, and K. B. Reed, “Developing a cerebral infarction,” Journal of Rehabilitation Research and gait enhancing mobile shoe to alter over-ground walking Development, vol. 32, pp. 236–244, 1995. coordination,” in 2012 IEEE International Conference on Robotics and Automation, pp. 4129–4142, Saint Paul, MN, [19] C. K. Balasubramanian, M. G. Bowden, R. R. Neptune, and USA, May 2012. S. A. Kautz, “Relationship between step length asymmetry and walking performance in subjects with chronic hemipar- [34] I. Handzic and K. B. Reed, “Kinetic shapes: analysis, verifica- esis,” Archives of Physical Medicine and Rehabilitation, tion, and applications,” Journal of Mechanical Design, vol. 88, no. 1, pp. 43–49, 2007. vol. 136, no. 6, article 061005, pp. 0610051–0610058, 2014. [20] M. G. Bowden, C. K. Balasubramanian, R. R. Neptune, and [35] R. C. Lynall, L. A. Zukowski, P. Plummer, and J. P. Mihalik, S. A. Kautz, “Anterior-posterior ground reaction forces as a “Reliability and validity of the protokinetics movement 8 Applied Bionics and Biomechanics analysis software in measuring center of pressure during walk- ing,” Gait & Posture, vol. 52, article S0966636216307111, pp. 308–311, 2017. [36] S. S. Ng and C. W. Hui-Chan, “The timed up & go test: its reliability and association with lower-limb impairments and locomotor capacities in people with chronic stroke,” Archives of Physical Medicine and Rehabilitation, vol. 86, no. 8, pp. 1641–1647, 2005. [37] P. S. Pohl, P. W. Duncan, S. Perera et al., “Influence of stroke- related impairments on performance in 6-minute walk test,” Journal of Rehabilitation Research and Development, vol. 39, no. 4, pp. 439–444, 2002. [38] T. D. V. Swinscow and M. J. Campbell, Statistics at square one, BMJ, London, 2002. [39] F. P. Carpes, C. B. Mota, and I. E. Faria, “On the bilateral asym- metry during running and cycling - A review considering leg preference,” Physical Therapy in Sport, vol. 11, no. 4, pp. 136–142, 2010. [40] I. Handzic, H. Muratagic, and K. Reed, “Passive kinematic syn- chronization of dissimilar and uncoupled rotating systems,” Nonlinear Dynamics and Systems Theory, vol. 15, pp. 383– 399, 2015. [41] H. Sadeghi, P. Allard, F. Prince, and H. Labelle, “Symmetry and limb dominance in able-bodied gait: a review,” Gait & Posture, vol. 12, no. 1, pp. 34–45, 2000. [42] M. J. Highsmith, J. T. Kahle, S. L. Carey et al., “Kinetic asym- metry in transfemoral amputees while performing sit to stand and stand to sit movements,” Gait & Posture, vol. 34, no. 1, pp. 86–91, 2011. International Journal of Advances in Rotating Machinery Multimedia Journal of The Scientific Journal of Engineering World Journal Sensors Hindawi Hindawi Publishing Corporation Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 http://www www.hindawi.com .hindawi.com V Volume 2018 olume 2013 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Journal of Control Science and Engineering Advances in Civil Engineering Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Submit your manuscripts at www.hindawi.com Journal of Journal of Electrical and Computer Robotics Engineering Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 VLSI Design Advances in OptoElectronics International Journal of Modelling & Aerospace International Journal of Simulation Navigation and in Engineering Engineering Observation Hindawi Hindawi Hindawi Hindawi Volume 2018 Volume 2018 Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com www.hindawi.com www.hindawi.com Volume 2018 International Journal of Active and Passive International Journal of Antennas and Advances in Chemical Engineering Propagation Electronic Components Shock and Vibration Acoustics and Vibration Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018

Journal

Applied Bionics and BiomechanicsHindawi Publishing Corporation

Published: Nov 11, 2019

References