No announcement yet.

Subject: BioNet: Problems in myoskeletal inverse dynamics

This topic is closed.
  • Filter
  • Time
  • Show
Clear All
new posts

  • Subject: BioNet: Problems in myoskeletal inverse dynamics

    Dear all,
    Although many participants have already said many things with
    which I wholly agree, I just want to make a few comments.
    Some of the discussiants, e.g. Prof Hatze and Paul Ostic seem to
    consider more sophisticated models as the solution to the inverse
    dynamics problem. To refine existing models to the highest
    possible degree is a worthy pursuit and can give very valuable
    results. Two problems remain, however.
    1) The problem of measurement accuracy. All conceivable
    movement recording systems have some measurement error.
    Especially when positions are to be differentiated to get
    accelerations these errors are amplified strongly. The problem of
    markers shifting with the skin remains difficult. It is hard to
    conceive a model which would eliminate these effects.
    Partial solutions: more accurate position recording systems, direct
    recording of acceleration and angular velocity with accelerometers
    and gyroscopes.
    2) The problem. that any model, simple or complex, has to be fitted
    to the individual subject. Much may be improved here, but all
    improvements need measurements, MRI, ultrasound etc etc. For
    any research a balance has to be sought between accuracy and
    investments in measurement time and apparatus.

    I was somewhat suprised by the remarks of Richard Baker. He
    suggests that we know very little of errors in inverse dynamics. I
    do not agree with that. If you have sensible estimates of the
    accuracy of your position recording system, your force plate and
    some idea of the relation between marker positions and the 'true'
    positions of joint axes etc. you can make a sensible estimate of
    the measurement errors. Alraedy in Cappozzo's pioneering paper (
    J. Biomech. 9: 35-43, 1976) error margins have been given for
    ankle, knee and hip moment.
    It is understandable that VICON Inc. does not advertize these
    errors, nor that many clinical engineers do not stress this point too
    much to their medical bosses, who have provided the funding, but it
    is not difficult to estimate them. In my opinion it is just good sense
    to give inverse dynamics results together with an estimate of the
    error. Nothing is perfect, an imprecise result is better than none,
    but it is wasted time to draw conclusions from minor dips or bips
    which are smaller than the noise.

    Finally, there is the subjective impression of elegance. In my
    opinion an elegant model is not the most complicated model that
    can be conceived. Such models have so many parameters, that
    their output is always 'in excellent agreement with experiment'. For
    me an elegant theory or model is one which explains a suprising
    amount of experiments with a minimum of complexity. Some
    models of McMahon or McNeil Alexander, in which running or
    jumping humans are modelled as point masses bouncing on a
    spring, have this elegance in a high degree. Am I weird in this
    preference? (Maybe it is just because my mathematics is rather

    ************************************************** *****
    At Hof
    Institute of Human Movement Sciences &
    Laboratory of Human Movement Analysis AZG
    University of Groningen
    A. Deusinglaan 1, room 769

    PO Box 196
    Tel: (31) 50 3632645
    Fax: (31) 50 3632751
    ************************************************** *******

    To unsubscribe send SIGNOFF BIOMCH-L to
    For information and archives: