Quantifying Segmental Contributions to Center-of-Mass Motion During Dynamic Continuous Support Surface Perturbations Using Simplified Estimation Models

How do features of dynamic postural stability change with age during quiet standing, gait, and obstacle crossing?

Previous research has reported mixed findings regarding age-related changes in dynamic postural stability, quantified by margin of stability (MOS), during gait. However, age-related changes in MOS may be better elicited by tasks imposing greater challenges to the postural control system. Older adults' MOS during obstacle crossing, a destabilizing task, has previously been characterized, although studies comparing MOS during this task between younger and older adults remain sparse. This study investigated age-related changes in dynamic postural stability during quiet standing, gait, and obstacle crossing. Participants aged 20-30 (n = 20), 60-69 (n = 18), 70-79 (n = 15), and 80+ (n = 7; not analyzed statistically) years old performed these tasks while whole-body motion was tracked using motion capture. MOS in each direction was estimated throughout each trial, and integrals, transient ranges, and trial minima were extracted (as applicable). MOS time series were also ensemble averaged across age groups. No age-related differences were identified for quiet standing or gait. However, obstacle crossing metrics revealed greater stability (i.e., more positive MOS) and less instability (i.e., less negative MOS) in older adults, and reduced ranges during transients. These findings potentially arise from shorter step lengths, which may be the result of age-related physical declines; or may reflect a cautious strategy in older adults, which maximizes postural stability in the direction with the greatest consequences for foot-obstacle contact, as it changes throughout the task. This study supports the use of tasks imposing physical challenges and/or voluntary perturbations to study age-related changes in dynamic postural stability. Findings also contribute to our theoretical understanding of the time course of dynamic postural stability during functional tasks in relation to periods of transition in the base of support, and task-specific strategies adopted for obstacle crossing by older adults to maintain dynamic postural stability and mitigate fall risk.

Influences of backpack loading on recovery from anterior and posterior losses of balance: An exploratory investigation

Backpacks are common devices for carrying external posterior loads. However, relatively little is known about how these external loads affect the ability to recover from balance loss. In this exploratory investigation, 16 young adults (8 female, 8 male) performed forward and backward lean-and-release balance recovery trials, while wearing a backpack that was unloaded or loaded (at 15% of individual body weight). We quantified the effects of backpack loading on balance recovery in terms of maximum recoverable lean angles, center-of-mass kinematics, and temporal-spatial stepping characteristics. Mean values of maximum lean angles were 20° and 9° in response to forward and backward perturbations, respectively. These angles significantly decreased when wearing the additional load for only backward losses of balance. During backward losses of balance, the additional load decreased peak center-of-mass velocity and increased acceleration by ∼10 and 18% respectively, which was accompanied by ∼5% faster stepping responses and steps that were ∼9% longer, 11% higher, and had an ∼10% earlier onset. Thus, wearing a backpack decreases backward balance recovery ability and changes backward recovery stepping characteristics.

A comparison of minimum segment models for the estimation of centre of mass position and velocity for slip recovery during a bathtub transfer task

BACKGROUND

Exploring the use of minimum marker sets is important for balancing the technical quality of motion capture with challenging data collection environments and protocols. While minimum marker sets have been demonstrated to be appropriate for evaluation of some motion patterns, there is limited evidence to support model choices for abrupt, asymmetrical, non-cyclic motion such as balance disturbance during a bathtub exit task.

RESEARCH QUESTION

How effective are six models of reduced complexity for the estimation of centre of mass (COM) displacement and velocity, relative to a full-body model.

METHODS

Eight participants completed a bathtub exit task. Participants received a balance perturbation as they crossed the bathtub rim, stepping from a soapy wet bathtub to a dry floor. Six reduced models were developed from the full, 72-marker, 12 segment 3D kinematic data set. Peak displacement and velocity of the body COM, and RMSE (relative to the full-body model) for displacement and velocity of the body COM were determined for each model.

RESULTS

Main effects were observed for peak right, left, anterior, posterior, upwards and downwards motion, and peak left, anterior, posterior, upwards and downwards velocity. Time-varying (RMSE) was smaller for models including the thighs than models not containing the thighs. In contrast, inclusion of upper arm, forearm, and hand segments did not improve model performance. The model containing the sacrum marker only consistently performed the worst across peak and RMSE metrics.

SIGNIFICANCE

Findings suggest a simplified centre of mass model may adequately capture abrupt, asymmetrical, non-cyclic tasks, such as balance disturbance recovery during obstacle crossing. A reduced kinematic model should include the thighs, trunk and pelvis segments, although models that are more complex are recommended, depending on the metrics of interest.

Estimation of sagittal-plane whole-body angular momentum during perturbed and unperturbed gait using simplified body models

Human whole-body angular momentum (WBAM) during walking typically follows a consistent pattern, making it a valuable indicator of the state of balance. However, calculating WBAM is labor-intensive, where the kinematic data for all body segments is needed, that is, based on a full-body model. In this study, we focused on selecting appropriate segments for estimating sagittal-plane WBAM during both unperturbed and perturbed gaits, which were segments with significant angular momentum contributions. Those major segments were constructed as a simplified model, and the sagittal-plane WBAM based on a simplified model was calculated by combining the angular momenta of the selected segments. We found that the WBAM estimated by seven-segment models, incorporating the head & torso (HT) and all lower limb segments, provided an average correlation coefficient of 0.99 and relative angular momentum percentage of 96.8% and exhibited the most similar sensitivity to external perturbations compared to the full-body model-based WBAM. Additionally, our findings revealed that the rotational angular momenta (RAM) of lower limb segments were much smaller than their translational angular momenta (TAM). The pair-wise comparisons between simplified models with and without RAMs of lower body segments were observed with no significant difference, indicating that RAMs of lower body segments are neglectable. This may further simplify the WBAM estimation based on the seven-segment model, eliminating the need to estimate the angular velocities of lower limb segments. These findings have practical implications for future studies of using inertial measurement units (IMUs) for estimating WBAM, as our results can help reduce the number of required sensors and simplify kinematics measurement.

Novel applications of Bayesian inference clarify sensorimotor uncertainty during stepping movements

The human nervous system relies on sensory information from the feet and legs to control the way we balance and walk. However, even in healthy individuals this sensory information is inherently variable and clouded with uncertainty. Researchers have found that the central nervous system (CNS) estimates body position amid the uncertainty of sensory signals in a way consistent with Bayesian inference. Bayesian inference posits that the brain accounts for variability in sensory data by combining it with learned expectations built from previous movement attempts. While initial findings on this topic are promising, they have neglected to study full-body movements such as gait and balance. The purpose of this research was to determine if the CNS controls balance-related stepping tasks in a way that fits a Bayesian framework. To address this purpose, we created a virtual reality protocol where participants moved their center of mass (CoM) to various targets while relying on uncertain visual cues and compensating for an alternating shift to the cursor position. We showed that as incoming sensory information became less certain, participants relied more on their learned expectation of body position and demonstrated more uncertainty in their responses. Accordingly, as participants learned to control and estimate their CoM position during our mobility task, they relied both on the sensory information they were receiving as well as learned expectations for its location. These results provide further evidence that the CNS is aware of the variability in sensory information and is proficient at compensating for the resultant uncertainty. We aim to apply these findings as a method for measuring the efficacy of interventions targeting sensory function.