Re: [Freesurfer] Longitudinal analysis--mass univariate {Disarmed}

[John Anderson]

Dear Dr Martin ,

Thanks a lot for your kindness and your great answers!!

Please I have one more question regarding the design matrix:

If I use my design matrix for LME depending on the exampels mentioned in FS wiki https://surfer.nmr.mgh.harvard.edu/fswiki/FsgdExamples

Is this correct?

 

 

Bests,
John 


 

Sent: Saturday, January 30, 2016 at 5:51 AM
From: "Martin Reuter" 
To: "Freesurfer support list" 
Subject: Re: [Freesurfer] Longitudinal analysis--mass univariate {Disarmed}


Hi John,
 

What I mean is there could be a bias in your study design if 99% of group 1 have three or more time points and  99% of group 2 have only two time points. This is something you should check. If distributions are approximately similar, you’d be fine. LME can be used in either case.  If there is any biases (also age, gender etc) you’d probably want to control for it via additional co-variates in your LME model. 

 

Best, Martin

 


On Jan 28, 2016, at 1:12 PM, John Anderson <j.ander...@publicist.com> wrote:
 




Thanks a lot D martin for your quick answer.

Regarding my patients I have two groups but the subject in every group have differnt time points. i.e Group-patients 1: I have 100 patients some scanned three times and some four times the rest at least two times.

Group-Patients 2:  is consisted of 75 patients some also some scanned three times and some four times the rest at least two times.

 

Is LME -- mas usnivariate still workable with out a bias ?

 

 

Bests,
John 


 

Sent: Thursday, January 28, 2016 at 6:51 AM
From: "Martin Reuter" <mreu...@nmr.mgh.harvard.edu>
To: "Freesurfer support list" <freesurfer@nmr.mgh.harvard.edu>
Subject: Re: [Freesurfer] Longitudinal analysis--mass univariate


Hi John,
 

you should check if the number of time points per subject is relatively random across your two patient groups. You don’t want a bias, let’s say one patient group with 2 tp and the other with 3.

 

You can compare the atrophy across these two groups easily with LME. Or you can also test if the rate is different from zero in each individual group. 

 

You will never be able to compute atrophy rates for your control group (as you only have a single time point for them) so you cannot compare with them either.

 

Best, Martin

 

 


On Jan 28, 2016, at 12:23 PM, John Anderson <MailScanner has detected a possible fraud attempt from "x-msg:" claiming to be j.ander...@publicist.com> wrote:
 




Hi Dr Martin,

I have two groups of patients and one group of controls.

The patients scanned multiple times but the number of time points is different between the subjects.

The controls have only one time point.

 

I aim to :

1. I wanted to study the changes in cortical thickness over time in each group of patients.

2. I wanted to compare the change in cortical thickness between patients and controls over time.

 

I followed wiki and I choose LME to run the analysis.

My questions are:

1. Can I use mass univariate analysis to check the changes in cortical thickness over time in only one group of patients. The example In wiki was for four groups.

2. Can I do a comparison in cortical thickness over time between controls ( who have  only one time point ) and patients using mass univariate approach.
 

Bests,
John 


___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer



___ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.






___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer



___ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.




___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.ed

Re: [Freesurfer] Longitudinal analysis--mass univariate {Disarmed}

[Martin Reuter]

Hi John,

What I mean is there could be a bias in your study design if 99% of group 1 
have three or more time points and  99% of group 2 have only two time points. 
This is something you should check. If distributions are approximately similar, 
you’d be fine. LME can be used in either case.  If there is any biases (also 
age, gender etc) you’d probably want to control for it via additional 
co-variates in your LME model. 

Best, Martin

> On Jan 28, 2016, at 1:12 PM, John Anderson  wrote:
> 
> Thanks a lot D martin for your quick answer.
> Regarding my patients I have two groups but the subject in every group have 
> differnt time points. i.e Group-patients 1: I have 100 patients some scanned 
> three times and some four times the rest at least two times.
> Group-Patients 2:  is consisted of 75 patients some also some scanned three 
> times and some four times the rest at least two times.
>  
> Is LME -- mas usnivariate still workable with out a bias ?
>  
>  
> Bests,
> John 
>  
> Sent: Thursday, January 28, 2016 at 6:51 AM
> From: "Martin Reuter" 
> To: "Freesurfer support list" 
> Subject: Re: [Freesurfer] Longitudinal analysis--mass univariate
> Hi John,
>  
> you should check if the number of time points per subject is relatively 
> random across your two patient groups. You don’t want a bias, let’s say one 
> patient group with 2 tp and the other with 3.
>  
> You can compare the atrophy across these two groups easily with LME. Or you 
> can also test if the rate is different from zero in each individual group. 
>  
> You will never be able to compute atrophy rates for your control group (as 
> you only have a single time point for them) so you cannot compare with them 
> either.
>  
> Best, Martin
>  
>  
> On Jan 28, 2016, at 12:23 PM, John Anderson  > wrote:
>  
> Hi Dr Martin,
> I have two groups of patients and one group of controls.
> The patients scanned multiple times but the number of time points is 
> different between the subjects.
> The controls have only one time point.
>  
> I aim to :
> 1. I wanted to study the changes in cortical thickness over time in each 
> group of patients.
> 2. I wanted to compare the change in cortical thickness between patients and 
> controls over time.
>  
> I followed wiki and I choose LME to run the analysis.
> My questions are:
> 1. Can I use mass univariate analysis to check the changes in cortical 
> thickness over time in only one group of patients. The example In wiki was 
> for four groups.
> 2. Can I do a comparison in cortical thickness over time between controls ( 
> who have  only one time point ) and patients using mass univariate approach.
>  
> Bests,
> John 
> ___
> Freesurfer mailing list
> Freesurfer@nmr.mgh.harvard.edu 
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer 
> ___
>  Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu 
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer 
>  The 
> information in this e-mail is intended only for the person to whom it is 
> addressed. If you believe this e-mail was sent to you in error and the e-mail 
> contains patient information, please contact the Partners Compliance HelpLine 
> at http://www.partners.org/complianceline 
>  . If the e-mail was sent to you in 
> error but does not contain patient information, please contact the sender and 
> properly dispose of the e-mail.
> ___
> Freesurfer mailing list
> Freesurfer@nmr.mgh.harvard.edu
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.