The Fourth IEEE RAS/EMBS International Conference on Biomedical Robotics and Biomechatronics Roma, Italy. June 24-27, 2012 Characterization of Spatio-Temporal Parameters of Human Gait Assisted by a Robotic Walker Anselmo Frizera, Member, IEEE, Arlindo Elias, Antonio J. del-ama, Ramon Ceres and Teodiano Freire Bastos Abstract Investigations of the biomechanical parameters of robotic walker assisted gait are needed to allow modern rehabilitation strategies and to improve further technological developments. In this study, spatio-temporal gait parameters were assessed during normal and assisted ambulation with the Simbiosis walker model, a robotic walker with forearm supports. Six infra-red video cameras, integrated in a movement analysis system, were used for three-dimensional reconstruction of body segments and measurement of biomechanical variables during assisted ambulation. Results showed that walker-assisted gait was marked by an overall reduction of spatio-temporal parameters, especially gait speed, without modification in cadence-speed and stride length-speed relationships. Future work investigating such modality of assisted gait in clinical conditions are warranted and may contribute for a better understanding of user-device interaction forces and its impact over gait biomechanics. I. INTRODUCTION Spatio-temporal parameters are a basic pre-requisite for the complete analysis of human gait [1]. Variations of such parameters are observed in different clinical conditions, as well as within groups of healthy individuals [2]. The knowledge of how these parameters are affected by pathological or environmental factors are essential for the development of novel assistive devices to help in the design of modern rehabilitation strategies. Rollator-type walkers are important examples of assistive technology, which aim to empower the mobility of impaired individuals [3]. Due to its versatility and rehabilitation potential, the popularity of these models has increased, not only among people with restricted mobility, but also among research groups currently working in robotic versions of the device [4] [5]. However, when considering both conventional and robotic rollators, there are few studies in literature that address the biomechanical events that take place during this modality of assisted ambulation. Variations within walker models and implemented systems further complicate the analysis of the The Simbiosis Project is placed in the framework of the Spanish National Program of R&D Anselmo Frizera and Teodiano Freire Bastos are with Department of Electrical Engineering, Federal University of Espirito Santo - UFES, 512, Vitoria-ES, Brazil. anselmo@ele.ufes.br teodiano@ele.ufes.br Arlindo Elias is with the Department of Biotechnology, Federal University of Espirito Santo - UFES, 512, Vitoria-ES, Brazil. arlindofisio@yahoo.com.br Antonio J. del-ama is with the Spinal Cord Injury Hospital of Toledo, Spain. ajdela@sescam.jccm.es Ramon Ceres is with the Bioengineering Group, Consejo Superior de Investigaciones Cientificas (CSIC), Spain. ceres@iai.csic.es impact of such technology in user-device interactions and spatio-temporal gait parameters. Thus, for the prescription of the most adequate robotic model in clinical or domiciliary settings, a thorough analysis of the motion patterns that occur during assisted ambulation are needed. Moreover, the ways in which the action of walking with the device modifies the pathological gait patterns of impaired subjects must also be assessed. In this paper, healthy volunteers were enrolled in a pilot investigation of gait biomechanics during normal and walkerassisted ambulation using the Simbiosis model, a robotic walker with forearm supports [6]. The objective was to obtain reference standard parameters of walker-assisted gait [7]. After an initial discussion of the experimental protocol and the addressed parameters, the analysis of assisted ambulation, considering the spatio-temporal gait parameters, is presented. A. Participants II. METHODS Seven healthy male, age and height-matched, with no history of gait dysfunction, volunteered to participate in this pilot study. The subjects were not familiarized with the device and no previous test sessions were performed. Written informed consent for publication was obtained from all individuals. The Simbiosis Project was approved, along with the experimental and validation procedures, by Spanish Ministry of Science and Innovation (MICINN). B. Experimental Setting For the experiments conducted in this study, the BTS Bioengineering movement analysis system was used. Six infra-red video cameras were used for 3D reconstruction of the position of reflexive markers, placed over the subject s body according to the protocol described by Davis et al. [8]. Combining such information with the biomechanical model and subject s anthropometric data, it was possible to obtain the complete 3D motion of the subject s body segments during ambulation (Fig.1). The subjects were asked to walk in the laboratory walkway, while the video cameras tracked the attached markers during the trial. The subjects performed three different test sessions, which consisted of normal unassisted, slow unassisted and walker-assisted ambulation. The test procedure involving slow, unassisted ambulation was performed by five of the seven subjects. The objective was to observe if the relationship between speed, cadence and stride length were changed by the assisted gait modality. 978-1-4577-1198-5/12/$26.00 2012 IEEE 1087
Fig. 1. Interface of the BTS system showing the location of the reflexive markers and body segment reconstruction Due to markers occlusion, considering the experiments using the walker, the subjects were asked to repeat the tests, until nine complete and correct trials were recorded for each test condition. In the absence of accepted research guidelines for the walker model used in this study, a specific experimental protocol was designed, where no speed indication was provided for the subjects. This instruction was necessary to provide comfort, allow a natural speed during assisted gait and to avoid the user to force into other types of gait patterns. In the case of assisted-gait research, experimentation rules and procedures are not always clearly defined because of the different types of devices and variations within them. Specifically, in the field of walker-assisted gait, the vast majority of studies were performed with conventional type rollator walkers with handgrip supports. In such cases, handgrip height was previously adjusted to establish the experimentation standards [9]. In this study, the forearm supports height was adjusted as being the distance between the subject s elbow and the ground. The subjects were instructed to keep an upright trunk posture and 90 of elbow flexion while supported on the device at rest [9] [10] C. Spatio-Temporal Parameters The spatio-temporal parameters addressed in this study were (Fig. 2): Step Width (SW): Defined as the distance, in transverse plane, between both feet during walking (in meters). The medium point of both talocrural joints were used as reference point for this measure. Step Length (SL): Defined as the distance (in meters) between a specific point of one foot and the same point of the other foot. Stride Length (StrL): Distance, in the forward direction, between the initial contact of one foot and the next initial contact of the same foot. In this study, the test subjects did not present any pathological gait pattern so that it was possible to analyse only the stride length, since there were no significant differences between right and left step length. Cadence (CAD): Can be defined as the rhythm of a person s walk, usually expressed in steps per minute (steps/min). Fig. 2. Overview of the spatio-temporal parameters addressed Stance Phase Duration (St): Temporal interval between the initial contact of the calcaneus and the toe-off moment that marks the beginning of the oscillation phase. It is measured in time units or gait cycle percentage. Oscillation Phase Duration (Osc): Equally measured in time units or gait cycle percentage, this parameter is defined as the temporal interval between the toe-off and heel contact of the same foot. Double Support Duration (DS): Temporal interval in which both feet are simultaneously in contact with the ground during gait. It may also be expressed in time units or percentage of gait cycle. Spatial gait parameters were normalized according to the subject s height (h). Such normalization was necessary to allow comparisons between subjects with different heights [11]. Other spatio-temporal parameters, such as average speed, were also addressed and obtained directly from previously discussed measures (1). V avg[m/s] = D. Statistical Analysis StrL[m] CAD[steps/min] 120 Wilcoxon test was used to assess statistical differences between the measured parameters of the three target joints during normal and walker-assisted gait. The test was performed assuming a significance level of 0,05. Pearson coefficient (r) was used to assess correlation levels between unassisted and walker-assisted spatio-temporal parameters. III. RESULTS Table 1 presents the mean experimental values of spatiotemporal gait parameters of normal and walker-assisted gait. (1) 1088
Fig. 3. Cadence vs. Normalized gait speed according to subject s height Fig. 4. height Stride Length vs. Normalized gait speed according to subject s Previous studies [12] reported linear relationship between cadence and average gait speed (2). In this study, experimental results were provided to verify if the main spatiotemporal gait parameters were influenced by walking with the Simbiosis walker. Height adjusted data obtained by this procedure are presented in Fig. 3. overall experimental data. No significant differences were obtained for both plotted curves. Correlation coefficients (r = 0..852; r = 0.828; P <0.001) were also equivalent to literature data (r = 0.81; P <0.001) [15]. CAD[steps/min] = α 1 V avg[m/s] + α 0 (2) The relationship between speed, cadence, stride length, duration of stance and double support phases are well documented in literature [13] [14]. Considering that speed is a product of cadence by stride length, these three parameters are also strongly associated [15]. It was obtained, respectively for normal gait data and overall experiment data (unassisted and walker-assisted gait), α 1 = 112.826 and 110.918; and α 0 = 29.472 and 30.892. Correlation between normal gait data and overall experiment data showed no significant differences. Pearson correlation obtained were r = 0.978 and r = 0.974 for normal gait experiments overall experiment data, respectively (Fig. 3). Fig. 4 presents the relationship of normalized stride length and the average speed of the performed experiments with respective adjustments (logarithmic and linear), considering the data regarding normal gait and normal gait plus walkerassisted ambulation. StrL[h] = β 1 log(v avg[h/s]) + β 0 (3) Similarities were observed once more in the adjustments of the presented curves. For the logarithmic adjustments, it was obtained, respectively for normal gait data curve and for overall experimental data, β 0 = 0.878 and 0.886, and β 1 = 0.267 and 0.280. On the other hand, considering the linear adjustments, it was obtained respectively α 0 = 0.448 and 0.449, and α 1 = 0.474 and 0.473, respectively. High Pearson correlation coefficients were obtained for both normal gait experiments (r = 0.936) and overall data (r = 0.927). Fig. 5 illustrates the comparison between cadence and stride length over the three experimental conditions. Curve adjustments were again provided for normal gait data and Fig. 5. Normalized Stride Length vs. Cadence Fig. 6 presents the relation between duration of stance and oscillation phases with cadence, in seconds and percentage of the gait cycle. Adjustments were performed taking into consideration normal gait with different speeds (slow or selfselected) and compared with linear adjustment of walkerassisted experiments. IV. DISCUSSION Biomechanical analysis of unassisted human gait is well discussed in literature. Although distinct protocols and acquisition systems are available, comparisons between different investigations groups can be performed because of the definition of standard test procedures. As observed in Table 1, walker-assisted spatial gait parameters were smaller than unassisted ambulation at a selfselected speed. In terms of temporal parameters, it was observed a reduction of oscillation phase, followed by an increase of the duration of stance and double support phases. 1089
TABLE I SPATIO-TEMPORAL PARAMETERS OF UNASSISTED AND WALKER-ASSISTED GAIT. AVERAGE VALUES Subject Gait Assistance SW StrL CAD StP right StP left OscP right OscP left DSP [m] [m] [steps/min] [%cycle] [%cycle] [%cycle] [%cycle] [%cycle] 1 Normal 0.094 0.790 106.758 61.311 59.767 39.567 40.522 20.200 Walker-assisted 0.088 0.697 84.371 63.011 62.422 37.956 38.989 24.389 2 Normal 0.090 0.745 109.958 59.656 59.244 38.356 38.867 20.967 Walker-assisted 0.085 0.666 80.453 59.856 60.567 39.163 38.556 21.656 3 Normal 0.083 0.807 103.166 59.800 61.580 39.940 37.820 21.300 Walker-assisted 0.075 0.712 91.381 61.543 63.971 37.700 36.186 25.371 4 Normal 0.121 0.799 105.036 62.160 62.560 38.320 37.760 24.380 Walker-assisted 0.097 0.679 91.748 63.160 62.680 35.550 35.840 27.320 5 Normal 0.092 0.756 101.014 61.286 60.657 38.986 38.686 22.000 Walker-assisted 0.080 0.691 80.706 61.589 60.667 38.567 39.963 22.844 6 Normal 0.084 0.773 112.030 57.711 59.400 41.722 40.000 17.744 Walker-assisted 0.081 0.668 76.426 59.429 62.471 38.957 39.586 21.271 7 Normal 0.085 0.707 101.237 60.433 61.744 39.589 38.867 21.744 Walker-assisted 0.074 0.642 82.094 62.856 63.111 36.689 36.800 26.322 Normal 0.093 0.768 105.600 60.337 60.708 39.497 38.932 21.191 Mean 0.013 0.035 4.242 1.462 1.291 1.163 1.029 2.002 Walker-assisted 0.083 0.679 83.883 61.635 62.270 37.813 37.560 24.168 0.008 0.023 5.759 1.514 1.245 1.267 1.467 2.328 Relation Normal/W-assisted 89.37 88.42 79.43 102.15 102.57 95.74 96.48 114.05 Fig. 6. Relationship between stance and duration gait phases with normalized average speed The results also indicate that healthy subjects, walking with the Simbiosis walker model, did not present changes in the relationship between gait speed and cadence. Previous studies showed that the relationship of stride length and average speed is marked by a logarithmic variation, as seen in equation (3). This relation can be approximated with a straight line when limited to particular speed bands [16], which has a similar equation to the one presented previously (3). A certain dispersion of data, in relation to the adjusted curve, can also be observed in figure 3 and corroborate previous findings of related studies [12]. Significant correlations (P 0.001) were obtained for the presented data, further corroborating previous results. In the study of Kirtley et al [15], Pearson correlation coefficients of 0.95 (P <0.001) were also obtained for the relation between average gait speed and cadence or stride length. No significant differences were observed between assisted and unassisted ambulation data. This result indicated that the use of the device by healthy subjects did not affect the relationship between cadence and stride length in the time of obtaining a given gait speed. Another interesting finding was the observation of a greater concentration of spatio-temporal gait parameters of normal and walker-assisted gait than the parameters obtained by slower ambulation (Figure 4). It is important to recall that no additional instructions regarding gait speed were provided for the subjects in any experiment. This finding corroborates the observation of Drillis et al [17], that stated that walking at a self-selected speed is marked by small variability of gait parameters (such as step time) when compared with slower or faster gait speeds. The results presented in this study indicates that the action of walking with Simbiosis walker assistance did not affect significantly the relationship between speed, cadence and stride length. In this way, it was possible to establish the basis for the study of duration of the stance and oscillation phases measured during the experimental procedure. Previous studies showed that stance phase duration and double support time (as a percentage of the gait cycle) are reduced when gait speed increases [16]. Double support phase are gradually reduced and finally disappears when the subject starts to run. On the contrary, the participation of the oscillation phase in the gait cycle (percentage) increases 1090
when the speed is increased. Stance and oscillation phases were also related with gait speed in a linear fashion by previous research [13] [12]. As expected, the results in this study showed that the stance phase time was reduced when speed increased during both unassisted and walker-assisted ambulation. Introduction of walker-assisted gait data did not affect significantly the linear adjustments to the relationship between oscillation phase and gait speed (Fig. 6). In this case, it can be seen in the inferior graphic that the oscillation time (in seconds) was reduced when the speed increased. The results indicate that the walker-assisted ambulation is slower but presented similar spatio-temporal parameters to normal gait at a self-selected speed. Based on the measured parameters, there is no evidence that the use of the Simbiosis device compromises the development of the user s gait. The similarities observed in the data of walker-assisted and normal gait, suggests that the robotic walker with forearm supports can be used as a gait training device in modern rehabilitation programs and domiciliary functional compensation strategies. V. CONCLUSIONS AND FUTURE WORKS A. Conclusions In this study, the analysis of spatio-temporal parameters during walker assisted ambulation were investigated. It was observed an overall reduction in gait speed during assisted ambulation, which did not affect significantly the cadencespeed and stride length-speed relationships. Furthermore, stance and oscillation phases presented expected values for the observed gait speed. Since few studies of walker-assisted biomechanics are available in literature, especially regarding robotic fourwheeled walkers with forearm supports, the results presented in this study are innovative in this research field and may contribute for the definition of normative gait parameters database for such models. [4] A. Abellanas, A. Frizera, R. Ceres, and R. Raya, Assessment of the laterality effects through forearm reaction forces in walker assisted gait, Procedia Chemistry, vol. 1, no. 1, pp. 1227 1230, Sep. 2009. [5] A. Elias, A. Frizera, and T. F. B. Filho, Robotic walkers from a clinical point of view: feature-based classification and proposal of a model for rehabilitation programs, in XIV Reunion de Trabajo en Procesamiento de la Informacion y Control - RPIC 2011, 2011, pp. 1 5. [6] A. Frizera, R. Ceres, J. L. Pons, A. Abellanas, and R. Raya, The Smart Walkers as Geriatric Assistive Device. The SIMBIOSIS Purpose. in Proceedings of the 6th International Conference of the International Society for Gerontechnology, 2008, pp. 1 6. [7] F. Prince, H. Corriveau, R. Hkbert, and D. A. Winter, Gait in the elderly, Gait and Posture, vol. 5, pp. 128 135, 1997. [8] R. B. Davis, S. Ounpuu, D. Tyburski, and J. R. Gage, A gait analysis data collection and reduction technique, Hum Movement Sci, vol. 10, no. 3, pp. 575 587, Mar. 1991. [9] T. Alkjaer, P. K. Larsen, G. Pedersen, L. H. Nielsen, and E. B. Simonsen, Biomechanical analysis of rollator walking. Biomedical engineering online, vol. 5, no. 2, Jan. 2006. [10] J. W. Youdas, B. J. Kotajarvi, D. J. Padgett, and K. R. Kaufman, Partial weight-bearing gait using conventional assistive devices. Archives of physical medicine and rehabilitation, vol. 86, no. 3, pp. 394 8, Mar. 2005. [11] A. L. Hof, Scaling gait data to body size, Gait and Posture, vol. 5, pp. 222 223, 1996. [12] C. Kirtley, Clinical Gait Analysis: Theory and Practice. Churchill Livingstone, 2006. [13] L. Larsson, P. Odenrick, B. Sandlund, P. Weitz, and P. Oberg, The phases of the stride and their interaction in human gait. Scand. J. Rehab. Med., vol. 12, pp. 107 112, 1980. [14] M. Murray, Gait as a total pattern of movement, Amer J Phys Med, vol. 46, pp. 290 333, 1967. [15] C. Kirtley, M. Whittle, and R. Jefferson, Influence of walking speed on gait parameters, Journal of Biomedical Engineering, vol. 7, no. 4, pp. 282 288, 1985. [16] D. Winter, Biomechanics and motor control of human movement, 2nd ed. John Wiley & Sons, Inc., 1990. [17] R. Drillis, Objective recording and biomechanics of pathological gait, Ann New York Acad Sci, vol. 17, pp. 86 109, 1958. B. Future Works Future investigations should focus on the gait analysis of impaired individuals and the impact of the walker assistance over biomechanical parameters. The data obtained in this study may be used as standard parameters for prospective comparisons in clinical investigations. Additionally, the costbenefits of this technology and improvements on the overall quality of life of the users should also be explored. REFERENCES [1] W. Zijlstra and A. L. Hof, Assessment of spatio-temporal gait parameters from trunk accelerations during human walking. Gait & posture, vol. 18, no. 2, pp. 1 10, Oct. 2003. [2] P. Esser, H. Dawes, J. Collett, M. G. Feltham, and K. Howells, Assessment of spatio-temporal gait parameters using inertial measurement units in neurological populations. Gait & posture, vol. 34, no. 4, pp. 558 60, Oct. 2011. [3] S. Stowe, J. Hopes, and G. Mulley, Gerotechnology series: 2. Walking aids, European Geriatric Medicine, vol. 1, no. 2, pp. 122 127, May 2010. 1091