Accepted Version

Version of code corresponding to manuscript acceptance at NeuroImage: Clinical.

NeuroImage: Clinical submission version

Release version used for manuscript submission to NeuroImage: Clinical. Tests three tracts for differences between low and high PARS-6 groups using a General Additive Model. FA values from nodes which differ in fit estimations are then averaged and used to test for a relation with negative lure generalization.

Pre-pairwise comparisons

This version accepts group type 1 (Con v Anx), 2 (Con v GAD v SAD), or 3 (L v M v H PARS6) and either does pairwise (group 1) or anova (group 2, 3) comparisons. For differences in splines on group 2, 3 only the nodes differing between all groups are used. The next version will remove group type 1, and only compare Con v SAD and L v H.

Pre-group 3

Refactored state before adding a third (PARS6) group.

Pre-refactoring

State of project before major refactoring of afq_step3_stats.R.

First Analysis

First set of analyses. GAM uses three anx groups (PDS low, mid, high), and then node with max group difference in predicted FA is used to predict LG/DI scores.

First Pass Success

Pipeline of AFQ through analyses working for multiple tracts.

1. AFQ
2. GAM
3. Differences in splines
4. Regress for behavior.