ECP Update
5/26/21
Stephen
New Pre-processed data
- Did the group average matrices change?
- Are nearest neighbor classifiers any better?
- The average correlation between a HC's previous connectivity matrix and the new one is .8735
- The correlation between the average HC matrices is .9784
Repeat Last Finding
- Repeated statistic analysis from previous paper
- Two TLE subjects didn't make the cuttoff in Joe's new pipeline :(
The result was no change.
$$r^2=.95$$
Nature Protocol
- Just followed the figure
- Used F-test on each edge for 4 categories (HC, L,R, Bi)
- Converted F-test to p value
pValue Distribution
Result
Notes
- Unable to do "CWAS" type as don't have patient drug info
Cameron's T-maps
- One of the questions of regarding the use of a mask is how consistent are the masks across folds?
- I used 5 fold cross validation on the new data. Used 1,000 permutations per fold to estimate corrected t-values
- Then visualized on each node by summing t-values
Left
Relative to
Control
1
2
3
4
5
