View Full Version : Accelerometer Validation: replies

Jason Purton
04-09-2001, 08:58 PM
I would like to thank everyone for their information. Sorry about the
delay, but here are the postings to my question about accelerometers and
balance and under them are the replies on the validation of accelerometers.

Original Question:

I was considering the use of 3D accelerometers in the assessment of
postural stability and was wondering does anyone know of any research that
validates (validity, reliability, ICC, etc) the use of accelerometers
against other devices used for the analysis of postural stability (eg.
Force platforms, Video analysis) or clinical balance tests. I have
articles on the assessment of accelerometers in assessing acceleration but
nothing against other devices. A summary post of the replies will be

Reply 1

There is a new review article related to this topic in Exercise and Sport
Sciences Reviews, V29, p 32-36, 2001. The title is, "Objective measurement
of physical activity in youth: current issues, future directions.," by
Stewart Trost.
The review reports evaluations of the Tri-Trac 3D accelerometer and one or
two 1D accelerometers. These devices are usually used to assess locomotion
activity and not postural sway. I do not know if they are sensitive enough
to measure such low level accelerations.
(NOTE: this Article was more on the assessment of energy expenditure than
on balance assessment)

Reply 2

I am not aware of any such study. The nearest paper would be:
"Detection of posture and motion by accelerometry: a validation study in
ambulatory monitoring"
F. Foerster, M. Smeja and J. Fahrenberg
Computers in Human Behavior 15, (1999) 571-583

Reply 3
There is a technical note where the triaxial accelerometers were compared
to a conventional analysis using simultaneously recorded kinematics and
ground reaction forces, but during the single phase support of walking and
running not for the analysis of postural stability. Anyway maybe it can
help you.
J Biomech 1996 Jul;29(7):949-54
A method for inverse dynamic analysis using accelerometry.
Van den Bogert A, Read L, Nigg BM

Reply 4
assessment of postural stability? Are these movements with small velocities
and accelerations?
The problem with accelerometers in principal is that they are bad in
measuring movements with very low accelerations (quasi-static situations).
All devices have an upper critical frequency but accelerometers have an
lower critical frequency in addition.
If you are interested in absolute values of position (or velocity) over a
longer time you may come in trouble even if the movement is characterised
with higher accelerations. The position is drifting away.
Please excuse me if I try to explain things that you are already know.

Secondary Question
It seems that there is work being done on the validation of accelerometers
against force platforms and video but there is no published research on the
area. Many of the replies were asking for further information on how to
assess an accelerometer's accuracy. This can only be done against a device
of known acceleration. Rather than list the methods that I have found I
will put the question to those developing and testing the devices.
Q. What are the better methods of testing an accelerometers accuracy.

Reply 1
I am not aware of any such study. The nearest paper would be:
"Detection of posture and motion by accelerometry: a validation study in
ambulatory monitoring"
F. Foerster, M. Smeja and J. Fahrenberg
Computers in Human Behavior 15, (1999) 571-583

Reply 2
For DC sensitive accelerometers a simple and precise calibration is to
measure the acceleration of gravity. Just position it upright, on its side,
and upside down, to record +g, 0 , -g. Any more calibration requires very
expensive equipment, a vibrator and a calibrated accelerometer. Probably
there are labs that can do it for you. The Danish firm Bruel & Kjaer is the
leading manufacturer in this field.
That there is no published research on accelerometry in
biomechanics is not true. Our list moderator Ton van den Bogert
has done some nice work in this field. Other work is from Ge Wu of
Penn State USA. Just look trough the last few years of J. Biomech
and Med Biol Engg

Reply 3
Tests for an accelerometer depend on the type and use of the device.
A simple test static response of common low-range accelerometers (say up to
+/- 3g) uses earth gravity as 'a device of known acceleration'. Allows
checking values of -1g...+1g, and axis alignment. Off course, this test is
not very usefull for sensors which cannot measure down to DC (0 Hz).
For dynamic behaviour and other test, technical specifications can often be
used. If you really need to check these specifications, a calibrated
'acceleration platforms' can be used to perform a lot of other, dynamic
tests, at a range of frequencies and amplitudes. Tests include e.g. cross
sensitivity for acceleration perpendicular on sensitive axis, linearity,
hysteresis etc.
For some purposes, a swing or a mass/spring system can be used to improvise
an 'acceleration platform' with limited possibilities.

Smile it makes people wonder what you are up to.

Jason Purton BHMS (Hons)
Staffordshire University
+44 01782 298508

To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l