Page 31 - Mechatronics for Safety Security and Dependability in a New Era

Ch04-I044963.fm Page 14 Tuesday, August 1, 2006 6:33 PM
Ch04-I044963.fm
14 14 Page 14 Tuesday, August 1, 2006 6:33 PM

Although falling is a result of complex and multi-factorial problem, lack of postural control is one
of the major contributing factors. Aging effects on the sensory feedback have been hypothesized
to be a key factor in adjusting posture to maintain their balance against unpredictable external
or internal variations of gait. In addition, recent randomized controlled trials that have tested
the effectiveness of the intervention for elderly have indicated that exercise training significantly
increase their aerobic capacity and muscle strength, which might result in improvement of the
postural stability.

Conventionally, clinicians have been assessed personnel walking ability based on performance of
static balance tests and measure of simple gait factors (walking speed, cadence, step length, etc.),
mostly focusing on quantifying regional amount of body sway, variability of gait factors or joint
angles. Those methods provide a practical evaluation, however, the measure of static balance
or gait variability itself does not mean that of dynamic stability of walking. Dynamic stability
represents a resilient ability to maintain certain continuous cyclic movement by accommodating
internal or external perturbations (Hurmuzlu et al. 1994). On the other hand variety of instruments
have been used to quantify walking characteristics in a more precise manner, by means of the
video-based motion capture system, goniometry, or force plates. However, those methods requires
considerable setups, then limited to laboratorial environments. Recently, mechatronics progress
made it possible to realize small and low power consumptive accelerometry as a testing tool
applicable in the field of medical therapy(Aminian et al. 2002, Ohtaki et al. 2001, 2005). Some
advanced algorithms have been also proposed to evaluate gait performances and dynamic walking
stability basing on a simple accelerometry(Dingwell et al. 2000, 2001, Buzzi et al. 2003. Arif
et al. 2004). Further application of those method to a physical assessment is strongly required
to enhance efficiency and effectiveness of interventions. Nevertheless, it is still insufficient to
investigate subsequent improvements on walking abilities in terms of the stability of dynamical
system.

This study was intended to present a practical method to assess walking stabilit}' by using a
portable instrument, then to investigate its usefulness in the physical assessment for elderly peo-
ple. The method employed a measurement of three-dimensional acceleration of the body, and
an application of nonlinear time-series analysis which directly assess stability of the dynamical
system. Straight level-walking of young and elderly subjects were investigated in the experiment.
Moreover, its feasibility in assessing effects and efficacies of the five-month interventions including
aerobic exercise training was investigated.

METHODOLOGY

In this study, we focused on local dynamic stability which is defined as a. sensitivity of the dynam-
ical system to small perturbations in gait variability which produced by one's locomotor system
itself. Lyapunov exponent estimation was applied to evaluate the local dynamic stability of walk-
ing. Firstly, state space was reconstructed from the obtained acceleration data after determining
appropriate time delay and embedding dimension:

y(t) = (x(t),x(t + r),---,x(t + {d- l)r)). (1)
Where, y(t) is the d dimensional state vector, x{t) is the original acceleration data, r is the time
delay, and d is the embedding dimension. A Schematic representation of the reconstruction process
was shown in Figure 1. A valid state space must include a sufficient number of coordinates to
unequivocally define the state of the attractor trajectories. Time delay r was determined as a
time when autocorrelation coefficient of the data gets lower than the reciprocal value of natural