Using Sensors and Deep Learning to Enable On-Demand Balance Evaluation for Effective Physical Therapy

As the center of mass (CoM) position can be used to determine stability, current rehabilitation standards may be improved by tracking it. A personalized CoM estimate can be obtained using the statically equivalent serial chain (SESC) once the model parameters are identified. The identification phase can be completed using low-cost sensors (Kinect and WBB) outside the laboratory making CoM estimation feasible in a patient’s home. This work focuses on: (1) improving the SESC identification quality and speed, and (2) using the estimated CoM to determine stability. Identification time is reduced by creating a visual adaptive interface where the subject’s limbs are colored based on the convergence of the SESC parameters. A study was conducted on eight subjects and showed a faster convergence and lower root mean square error (rmse) when the adaptive interface was used. We found that a model capable of estimating the CoM position with an rmse of 27 mm could be obtained after only 90 s of identification when the interface was used, whereas twice as much time was needed when the interface was not used. The interface that was developed can be used by a subject to track his/her CoM position in a self-directed way. Stability was determined for a squat task using a dynamic index obtained from the estimated CoM trajectory and using only Kinect measurements. This shows one potential application for home rehabilitation and monitoring.

102. An introductory investigation of standing postural control ‘error’ and its relation to balance in adult spinal deformity surgery patients

The Inverted Pendulum model is one of favorite algorithm used to calculate the length of gait. The human locomotion is approximated by simple pendulum, the influence of knee and ankle movements are not considered. The gait length is calculated from the extremes of vertical position of human Center of Mass (CoM) during gait. This approach is usually used in measurements which uses inertial sensors to measure the position of CoM. The position of CoM is after that calculated usually as double integration of acceleration. This approach causes errors in calculation, which are given by method (integration) and also by tilt of inertial sensor during movement of CoM. The influence of the Inverted Pendulum approach is not usually considered. This article shows the influence of Inverted Pendulum model as a methodology to calculate the length of gait. There were done measurement of human walk by precise Video System.

We examined the association between changes in swimming velocity, vertical center of mass (CoM) position, and projected frontal area (PFA) during maximal 200-m front crawl. Three well-trained male swimmers performed a single maximal 200-m front crawl in an indoor 25-m pool. Three-dimensional (3D) shape data of the whole body were fitted to 3D motion data during swimming by using inverse kinematics computation to estimate PFA accurately. Swimming velocity decreased, the vertical CoM position was lowered, and PFA increased with swimming distance. There were significant correlations between swimming velocity and vertical CoM position (|r| = 0.797–0.982) and between swimming velocity and PFA (|r| = 0.716–0.884) for each swimmer. These results suggest that descent of the swimmer’s body and increasing PFA with swimming distance are associated with decreasing swimming velocity, although the causal factor remains unclear.

When we face a super-aging society, there is a drastically increased need for efficient systems in terms of time and cost that can improve rehabilitation standards for the elderly people and other motor-impaired subjects. Human balance ability depends largely on the control of the full body center of mass (CoM), fall risks can be evaluated by estimating the subject-specific CoM displacement over the support polygon relating to the foot tracking. The CoM position is often estimated using what are known as anthropometric tables in biomechanics field. However, the parameters have been obtained from a standard population and will differ between subjects. Current existing fall risk assessment usually relies on the generic anthropometric table or need the center of pressure (CoP) recordings which are to be evaluated for the risk of fall. CoP measurements require force platform, wii board or specialized shoes, which limits the practical usage in the daily life environment. This work represents a personalized measure of balance that considers subject-specific body mass variations along with the motion tracking by Kinect Two. Based on our previous developments, we firstly verified the system with Kinect Two recording, and with adaptive support polygon extraction process, it realizes a real-time system for evaluating the personalized balance and fall risk visualization for unknown disturbance without needing force platform.

Purpose/Hypothesis: The purpose of this study is two fold: to investigate, in unperturbed walking, whether step width or frontal plane center of mass (COM) control can predict a person's ability to recover from an induced slip; and to investigate the extent to which frontal plane COM and BOS control during an anterior posterior (A-P) slip affect the outcome of the slip. We hypothesize that the step width and frontal plane COM control exhibited during natural gait and during the first, unexpected slip, will differentiate between those who will be able to recover from an induced slip and those who will not. Subjects: 34 healthy adults, 65 years and older. Materials/Methods: Subjects walked on an instrumented walkway while wearing a safety harness during motion capture of their regular gait. After 10 unperturbed trials, an AP slip was induced unexpectedly under the right heel at touchdown (TD). Motion data were used to calculate the frontal plane variables step width, COM position, normalized COM position, and COM velocity. These were calculated at right heel TD, left foot lift off (LO), and subsequent right foot touchdown (RCTD). Subjects were grouped by outcome of the first slip trial into the fall or the recover group. T-tests were used to compare the frontal plane variables at the three gait instants for the fall vs. the recover group for natural walking and for the slip trial. Results: Upon the first slip, all 34 subjects experienced a loss of balance: 19 subjects fell (fall group), the other 15 did not (recover group). During unperturbed walking, at TD and LO, there were no differences in the frontal plane variables between the fall group and the recover group. At left touchdown (RCTD), there were small, but significant differences in the lateral position of the COM and the step width between the two groups; but no differences in normalized COM position or COM velocity. At the three gait instants examined during the first, unexpected slip, there were no differences between the fall group and the recover group in step width, or any of the other frontal plane COM variables measured. Conclusions: Altered frontal plane center of mass control in stance, stepping, normal walking, and perturbed gait, as well as increased step width during gait, have been implicated as risk factors for falling associated with aging. We, therefore, expected step width and frontal plane COM control during natural walking and throughout a slip to be different between the group who fell as a result of the slip and those who were able to recover. We found only small differences during natural walking and none during the slip. Clinical Relevance: Balance and gait training in older adults which have the goal of fall prevention should focus on those areas which pre-dispose one to falling. In this case, step width and lateral control of the body's COM, during natural gait and during a slip, appear not to have been, as is frequently believed, related to the subject's ability to recover from an induced slip.

Center of mass estimation during non-cyclic activities: Comparison of marker-based methods and their fusion with ground reaction forces.

Control of center of mass motion state through cuing and decoupling of spontaneous gait parameters in level walking

ABSTRACT. The purpose of this study was to clarify the age-related effects of distances from the center of pressure (COP) to the center of mass (COM) (COP-COM distances) during one-leg standing (OLS) task. Healthy old and young adults (n = 11 each) participated in this study. The authors divided the task into 3 phases (accelerated, decelerated, and steady) based on the relationship between COM and COP. COP-COM distances in the older group were significantly reduced during the accelerated phase, then significantly increased during the decelerated and steady phases. Furthermore, distances during these phases correlated inversely with OLS time. The authors conclude that OLS time is shortened by the larger braking response to COM shifts just after leg-lifting, and the production of larger inertial forces to maintain COM position during the OLS in older individuals.

The effect of aging on vertical postural control during the forward and backward shift of the center of pressure

Since the center of mass (CoM) is related to ground reaction force (GRF), the CoM position is expected to be used in evaluation of walking ability. The purpose of this study was to examine the possibilities of using the CoM acceleration calculated from the CoM trajectory estimated by inertial sensor signals to obtain similar information as the GRF. In this paper, 16 m straight walking measurements were performed under different walking speeds with healthy subjects. CoM acceleration, which was calculated from CoM positions estimated by signals measured with inertial sensors, were analyzed for the steady state gait. Waveforms of the CoM acceleration were similar among subjects. Four peak values (negative peak near the toe off (TO) and positive peak near the initial contact (IC) of the opposite lower limb in the anteroposterior component, and the positive peak before the TO of the opposite lower limb and the negative peak of the mid-stance phase) showed significant correlations with walking speed, which were similar to the relationships between GRF and walking speed in previous studies. These results suggest that anteroposterior and vertical components of CoM acceleration estimated from inertial sensor signals have a possibility of estimating similar information as the GRF during walking. The relationship between positive peak value of the vertical component near the IC of the opposite side and walking speed was seemed to be different from a previous study, although it showed significant correlation with walking speed. Simultaneous measurements with inertial sensors and GRF are expected in the next study.

This work was supported by IT R & D program of MIC & IITA (2006-S-028-01, Development of Cooperative Network-based Humanoids Technology), Republic of Korea.

The complex structure of the human body makes its center of mass (CoM) estimation very challenging. The typically used estimation methods usually suffer from large estimation errors when applied to bodies with structural differences. Thus, a reliable estimation method is of utmost importance. In this paper, we present a detailed evaluation of a subject-specific CoM estimation technique named Statically Equivalent Serial Chain (SESC) by investigating its estimation ability over two different groups of subjects (Fit and Obese) in comparison to the segmental analysis method. For this study, we used an IMU-based motion capture system and a force platform to record the joint angles and corresponding center of pressure (CoP) values of twenty-five participants while performing a series of static postures. The root-mean-square errors (RMSE) of SESC’s estimation for both groups showed close and lower mean values, whereas the segmental analysis method showed significantly larger RMSE values in comparison to SESC (p < 0.05). In addition, we used the Bland–Altman analysis to evaluate the agreement between the two techniques and the ground truth CoP, which showed the accuracy, precision, and reliability of SESC over both groups. In contrast, the segmental analysis method did not present neither accurate nor precise estimations, as our analysis revealed considerable fixed and proportional biases.

Climbing animals theoretically should optimize the energetic costs of vertical climbing while also maintaining stability. Many modifications to climbing behaviors have been proposed as methods of satisfying these criteria, focusing on controlling the center of mass (COM) during ascent. However, the link between COM movements and metabolic energy costs has yet to be evaluated empirically. In this study, we manipulated climbing conditions across three experimental setups to elicit changes in COM position, and measured the impact of these changes upon metabolic costs across a sample of 14 humans. Metabolic energy was assessed via open flow respirometry, while COM movements were tracked both automatically and manually. Our findings demonstrate that, despite inducing variation in COM position, the energetic costs of climbing remained consistent across all three setups. Differences in energetic costs were similarly not affected by body mass; however, velocity had a significant impact upon both cost of transport and cost of locomotion, but such a relationship disappeared when accounting for metabolic costs per stride. These findings suggest that climbing has inescapable metabolic demands driven by gaining height, and that attempts to mitigate such a cost, with perhaps the exception of increasing speed, have only minimal impacts. We also demonstrate that metabolic and mechanical energy costs are largely uncorrelated. Collectively, we argue that these data refute the idea that efficient locomotion is the primary aim during climbing. Instead, adaptations towards effective climbing should focus on stability and reducing the risk of falling, as opposed to enhancing the metabolic efficiency of locomotion.

Athletes with non-contact anterior cruciate ligament tears have common features in the sagittal plane; namely, the body’s center of mass (COM) is located posterior to the base of support, the trunk and knee joints are extended, and the hip angle is flexed. However, the relationships among these variables have not been assessed in field-based movements. This study sought to determine relationships between distances from the COM to the base of support and the trunk, hip, and knee positions in women while playing soccer. Sixty events (29 single-leg landing and 31 single-leg stopping events) were analyzed using two-dimensional video analysis. The relationships among the measurement variables were determined using the Pearson’s product-moment correlation coefficient, and stepwise multiple linear regression models were used to explore the relationships between the COM position and the kinematic variables. The distance from the COM to the base of support displayed a moderate negative relationship with the trunk angle (r = −0.623, p < .0001, r2 = 0.388) and a strong positive relationship with the limb angle (r = 0.869, p < .0001, r2 = 0.755). The limb, knee, and trunk angles were selected in the best regression model (adjusted r2 = 0.953, p < .0001, f2 = 20.277). These findings suggest that an increased trunk angle and a decreased limb angle at initial contact are associated with a safer COM position. Neuromuscular training may be useful for controlling the trunk and lower limb positions during dynamic activities.