Back to PLS Help

Behavioural PLS and Multiple Brain Data Conditions
pdhami
Posted on 09/04/20 13:24:40
Number of posts: 36
pdhami posts:

Greetings PLS/Baycrest community,

I have a study in which participants were stimulated at six separate cortical sites while undergoing EEG recording. In this same group of participants, multiple behavioural/cognitive tests were adminstered. 

My question of interest is whether EEG data from perturbation of multiple cortical sites can better predict/correlate with the cognitive scores compared to perturbation of a single site only. From my readings, a multivariate analysis of my data might be best to incorporate the data from these six stimulation sites (which can be thought of as tasks conditions) together and relate them to behaviour.

Reading papers related to PLS (particularly this one https://www.sciencedirect.com/science/article/pii/S1053811910010074#bb0145), it seems as though behavioural PLS may be relevant to what I am trying to do. However, the text commonly refers to a single X matrix which makes me unsure if I have the right idea (in my head I'm thinking of six 'X' data matrices, one for each condition).

With my question in mind, is it something that can be investigated using PLS?

Thank you in advance for any comments or suggestions.

Best,

Paul

Replies:

Untitled Post
rmcintosh
Posted on 09/04/20 13:32:41
Number of posts: 385
rmcintosh replies:

quote:

Greetings PLS/Baycrest community,

I have a study in which participants were stimulated at six separate cortical sites while undergoing EEG recording. In this same group of participants, multiple behavioural/cognitive tests were adminstered. 

My question of interest is whether EEG data from perturbation of multiple cortical sites can better predict/correlate with the cognitive scores compared to perturbation of a single site only. From my readings, a multivariate analysis of my data might be best to incorporate the data from these six stimulation sites (which can be thought of as tasks conditions) together and relate them to behaviour.

Reading papers related to PLS (particularly this one https://www.sciencedirect.com/science/article/pii/S1053811910010074#bb0145), it seems as though behavioural PLS may be relevant to what I am trying to do. However, the text commonly refers to a single X matrix which makes me unsure if I have the right idea (in my head I'm thinking of six 'X' data matrices, one for each condition).

With my question in mind, is it something that can be investigated using PLS?

Thank you in advance for any comments or suggestions.

Best,

Paul

Hi Paul,

If I understand the experiment, then 'yes' PLS can be used for your question.  Each of the stimulation conditions would have a corresponding behavior (X) vector so you would comparing six correlation maps.

 

hope that helps

 

Randy



Untitled Post
pdhami
Posted on 09/04/20 18:16:00
Number of posts: 36
pdhami replies:

quote:

Hi Paul,

If I understand the experiment, then 'yes' PLS can be used for your question.  Each of the stimulation conditions would have a corresponding behavior (X) vector so you would comparing six correlation maps.

 

hope that helps

 

Randy

Dear Dr. McIntosh, 

thank you very much for the reply. 

Sorry for the follow up, but what I am hoping to address is this: using data from perturbation of X cortical sites (e.g. left and right DLPFC) better predicts cognitive performance than using data from perurbation of X - 1 cortical sites (e.g. just left DLPFC). I was thinking of using the 'merge conditions' option to potentially address this. Maybe I am not understanding PLS correctly in thinking if it can answer this question?

Thank you again,

Paul



Untitled Post
rmcintosh
Posted on 09/05/20 08:34:48
Number of posts: 385
rmcintosh replies:

quote:

Dear Dr. McIntosh, 

thank you very much for the reply. 

Sorry for the follow up, but what I am hoping to address is this: using data from perturbation of X cortical sites (e.g. left and right DLPFC) better predicts cognitive performance than using data from perurbation of X - 1 cortical sites (e.g. just left DLPFC). I was thinking of using the 'merge conditions' option to potentially address this. Maybe I am not understanding PLS correctly in thinking if it can answer this question?

Thank you again,

Paul

Hi Paul,

 

I am not sure that you can combine conditions as if the two sites were stimulated at the same time. If I understand your question, you want to see if stimulating left and right together yields better prediction of cogntion that each site alone.  From what you write, it seems you have data one for left stimulation and right stimulation, but not left and right together.  Without such data, I don't see how you can address your question. Am I missing something?



Untitled Post
pdhami
Posted on 09/05/20 09:36:30
Number of posts: 36
pdhami replies:

quote:

Hi Paul,

 

I am not sure that you can combine conditions as if the two sites were stimulated at the same time. If I understand your question, you want to see if stimulating left and right together yields better prediction of cogntion that each site alone.  From what you write, it seems you have data one for left stimulation and right stimulation, but not left and right together.  Without such data, I don't see how you can address your question. Am I missing something?

Hi Dr. McIntosh,

what you say is correct. Each site was stimulated independently (and not at the same time). For example, the left DLPFC would be stimulated while EEG was concurently recorded, and then the right DLPFC aftwards was stimulated with EEG concurently recorded, meaning there are two separate EEG files, one for left DLPFC, and one for right DLPFC.

I was hoping to see if a combination of the EEG data collected from multiple sites better predicts cognition than data alone from either of the six sites alone. But from what you mentioned before, using Behaviour PLS, it would compare the six cortical sites (and their correlation maps) to see if one stimulated site better predicts cognition than the remaining others. If I am understanding correctly, this would not address the question if data from two sites (e.g. left and right DLPFC) better predicts cognitive performance than data alone from either site (e.g. left or right DLPFC).

Sorry for any confusion, but your insight is greatly appreciated.

All the best,

Paul




Login to reply to this topic.

  • Keep in touch

Enter your email above to receive electronic messages from Baycrest, including invitations to programs and events, newsletters, updates and other communications.
You can unsubscribe at any time.
Please refer to our Privacy Policy or contact us for more details.

  • Follow us on social
  • Facebook
  • Instagram
  • Linkedin
  • Pinterest
  • Twitter
  • YouTube

Contact Us:

3560 Bathurst Street
Toronto, Ontario
Canada M6A 2E1
Phone: (416) 785-2500

Baycrest is an academic health sciences centre fully affiliated with the University of Toronto