Effects of external loads on balance control during upright stance: Experimental results and model-based predictions

Abstract

The purpose of this study was to identify the effects of external loads on balance control during upright stance, and to examine the ability of a new balance control model to predict these effects. External loads were applied to 12 young, healthy participants, and effects on balance control were characterized by center-of-pressure (COP) based measures. Several loading conditions were studied, involving combinations of load mass (10% and 20% of individual body mass) and height (at or 15% of stature above the whole-body COM). A balance control model based on an optimal control strategy was used to predict COP time series. It was assumed that a given individual would adopt the same neural optimal control mechanisms, identified in a no-load condition, under diverse external loading conditions. With the application of external loads, COP mean velocity in the anterior–posterior direction and RMS distance in the medial–lateral direction increased 8.1% and 10.4%, respectively. Predicted COP mean velocity and RMS distance in the anterior–posterior direction also increased with external loading, by 11.1% and 2.9%, respectively. Both experimental COP data and model-based predictions provided the same general conclusion, that application of larger external loads and loads more superior to the whole body center of mass lead to less effective postural control and perhaps a greater risk of loss of balance or falls. Thus, it can be concluded that the assumption about consistency in control mechanisms was partially supported, and it is the mechanical changes induced by external loads that primarily affect balance control.

Introduction

Falls are one of the most common incidents leading to injuries in daily activities and occupational settings. Fall-related injuries have substantial adverse impacts on functional ability and life quality. As unintentional falls often result from a ‘loss-of-balance’, an improved understanding of balance control may thus aid in understanding and preventing falls. A number of factors have been identified as influencing balance control, such as aging [1], [2], [3], localized muscle fatigue [4], [5], and decrements in the quality of sensory input [6].

External loads also appear to affect balance control, and many daily and occupational activities require load carriage (e.g. material handling and military marches [7]). Hence, further investigation of how and why balance control is affected by external loads is warranted. Previous studies have suggested that external loads adversely affect balance control, since such loads resulted in increased postural sway during quiet erect stance [7], [8], [9], [10]. Increased postural sway indicates that the whole-body center-of-mass (COM) gets closer to the limits of the base of support (BOS) and thus leads to less stability. Existing studies of external loads, however, have been somewhat narrowly focused on the effects of external load mass on balance control. The location of any external loads would also seem relevant; while this has been investigated with respect to energy costs [11], we are unaware of any evidence regarding whether load location affects balance control.

In addition to experimental studies, balance control models have been widely used to facilitate an understanding of underlying balance control mechanisms. For example, Ishida et al. [12] adopted a balance control model to identify the roles of different sensory systems, and Maurer and Peterka [13] applied a model based on a PID (proportional, integrative, and derivative) neural controller to examine the effects of aging on balance control. Evaluation of such mathematical models is often challenging. Some studies [13], [14] have examined whether balance control models could accurately simulate experimental center-of-pressure (COP) based measures, and Winter et al. [15] compared simulated and experimental relationships between sway amplitude and effective stiffness of balance control.

We recently presented a balance control model based on an optimal control strategy to simulate spontaneous sway behaviors, and used this model to identify aging effects on balance control [16]. Our model was shown to generate reasonable simulations of measures derived from COP time series. Models are expected to be able to address “what if” questions [17], thus one can argue that a stronger test involves determining whether this (or related) model can make predictions of sway or COP behaviors under novel circumstances. In the current context, these circumstances are different configurations of an external load during quiet upright stance.

One purpose of this study was to determine the actual effects of external loads on balance control, and specifically the influences of load mass and height, by using the data obtained while participants were loaded. We hypothesized that increasing mass and/or height would challenge balance control, based on expected mechanical effects for an analogous inverted pendulum (as is frequently used to model upright stance). The effects of external loads on balance control were identified by COP-based measures, since the COP reflects the net motor control signal output necessary to keep the projection of the center-of-mass (COM) within the BOS [18], [19]. Another purpose was to determine whether, or to what extent, our optimal control model could predict changes in balance control behaviors under novel conditions. Such ‘predictive ability’ was assessed using the same scenario involving application of external loads during quiet upright stance. This scenario was considered advantageous, since changes in mechanics due to loading could be estimated in a straightforward manner (see Section 2). Model predictions were based on an assumption that the same optimal control mechanisms, identified during unloaded conditions, are also employed during loaded conditions (see Section 2). Since experimental data were not used to generate behavioral simulations in the loaded conditions, we considered this a relatively strong test of the model's predictive ability.