search

brainhack-school2020 / stephaniealley_bhs2020_project

Effect of preprocessing on prediction performance of machine learning model
4 stars 0 forks source link

Considering the intentions behind confound removals #4

Open haqqeez opened 4 years ago

haqqeez commented 4 years ago

This project looks like it'll be tackling a very important issue of data standardization in science, good on you for taking the initiative on it!

I don't know much about MRI research, but I'm wondering if a developmental neuroscientist would selectively exclude a different set of confounds than say a neuroscientist studying sensory systems, memory, or mental illness? If so, I think it would be interesting to see how changing the kinds of confounds are included/excluded in an analysis could affect the overall interpretation of the results by the experimenter. It would also be interesting to consider when these exclusions are valid and when they might instead be biasing the final interpretation.

stephaniealley commented 4 years ago

That is a very interesting observation! In my own experience, I have found that it is not uncommon for the motivation behind such a choice to lack sufficient empirical justification and, thus, to bias the final interpretation. I will definitely keep this in mind for this project.