Tweet
Hello Colleagues,
I am a new user/reader in the Biomch-L list and have a question regarding
the application of bootstrap techniques to in vivo force measurement/gait
analysis data. Let me explain the data set I am dealing with and then ask
my question...
We have directly measured the in vivo forces (using implanted force
transducers) in a certain muscle-tendon unit of a quadruped (n=3 animals;
say N1, N2, N3) for different levels of activity (say A1, A2, A3). For each
activity level within an animal we have several repeated measurements of the
response variable, which may be different from each other, i.e.:
ACTIVITY A1 A2 A3
ANIMAL
N1 R11 R12 R13
N2 R21 R22 R33
N3 R31 R32 R33
where the Rij's (the number times a measurement is repeated) are not
necessarily equal. Further, we have three classes of response variables and
the ranges of Rij's for the three corresponding data sets are 8-10, 15-30,
and 1200-1500.
We are interested in finding out if changing the activity level (i.e.
subjecting the animals to activity A1 vs A2 vs A3) has an effect on the
response variables. Using the GLM (generalized linear model) procedure in
SAS, we performed ANOVA on our data and were able to state that changing
activity level significantly affects several of the response measures.
We submitted our data for publication and the study was received very well
by the reviewers. One of the reviewers, however, said that the statistical
analysis was inadequate because of a low sample size and suggested we review
and apply the bootstrap technique to evaluate our data.
Now, I only have a very rudimentary knowledge of the bootstrap and the Monte
Carlo techniques, all of which I have gleaned from reading through a section
on Bootstrapping and Monte Carlo Simulations in the "Numerical Recepies"
book. I did a literature review on bootstrap on medline and found only one
paper as applied to gait analysis (Sutherland et al, 1996) and was unable to
discern from it as to how to apply the technique to our data.
So here are the questions/requests I have:
1. How would it be useful to apply the bootstrap technique to analyze
our data? How will it help increase the confidence in our conclusions that
activity has a significant effect on various measures of in vivo force?
2. Can you guide me to a publication(s) which I can read through (and
perhaps further discuss with you) to help me formulate how to use this
technique?
Many thanks. As usual, I will summarize the replies I get.
I look forward to numerous replies (hopefully!)
Sincerely,
Prasanna Malaviya, Ph.D.
Georgia Institute of Technology
Atlanta, GA
(404) 894-2212, Fax: (404) 894-2291
prasanna.malaviya@me.gatech.edu
-------------------------------------------------------------------
To unsubscribe send UNSUBSCRIBE BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://www.bme.ccf.org/isb/biomch-l
-------------------------------------------------------------------
I am a new user/reader in the Biomch-L list and have a question regarding
the application of bootstrap techniques to in vivo force measurement/gait
analysis data. Let me explain the data set I am dealing with and then ask
my question...
We have directly measured the in vivo forces (using implanted force
transducers) in a certain muscle-tendon unit of a quadruped (n=3 animals;
say N1, N2, N3) for different levels of activity (say A1, A2, A3). For each
activity level within an animal we have several repeated measurements of the
response variable, which may be different from each other, i.e.:
ACTIVITY A1 A2 A3
ANIMAL
N1 R11 R12 R13
N2 R21 R22 R33
N3 R31 R32 R33
where the Rij's (the number times a measurement is repeated) are not
necessarily equal. Further, we have three classes of response variables and
the ranges of Rij's for the three corresponding data sets are 8-10, 15-30,
and 1200-1500.
We are interested in finding out if changing the activity level (i.e.
subjecting the animals to activity A1 vs A2 vs A3) has an effect on the
response variables. Using the GLM (generalized linear model) procedure in
SAS, we performed ANOVA on our data and were able to state that changing
activity level significantly affects several of the response measures.
We submitted our data for publication and the study was received very well
by the reviewers. One of the reviewers, however, said that the statistical
analysis was inadequate because of a low sample size and suggested we review
and apply the bootstrap technique to evaluate our data.
Now, I only have a very rudimentary knowledge of the bootstrap and the Monte
Carlo techniques, all of which I have gleaned from reading through a section
on Bootstrapping and Monte Carlo Simulations in the "Numerical Recepies"
book. I did a literature review on bootstrap on medline and found only one
paper as applied to gait analysis (Sutherland et al, 1996) and was unable to
discern from it as to how to apply the technique to our data.
So here are the questions/requests I have:
1. How would it be useful to apply the bootstrap technique to analyze
our data? How will it help increase the confidence in our conclusions that
activity has a significant effect on various measures of in vivo force?
2. Can you guide me to a publication(s) which I can read through (and
perhaps further discuss with you) to help me formulate how to use this
technique?
Many thanks. As usual, I will summarize the replies I get.
I look forward to numerous replies (hopefully!)
Sincerely,
Prasanna Malaviya, Ph.D.
Georgia Institute of Technology
Atlanta, GA
(404) 894-2212, Fax: (404) 894-2291
prasanna.malaviya@me.gatech.edu
-------------------------------------------------------------------
To unsubscribe send UNSUBSCRIBE BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://www.bme.ccf.org/isb/biomch-l
-------------------------------------------------------------------