Fig. 2: Univariate statistical analysis and univariate feature reduction hierarchical clustering analysis.

A, B Univariate analyses only reveal group differences in connectivity, and no regional change. A Regional power did not show any significant effects (correct for multiple comparisons), whereas B functional connectivity, for the beta band, shows only a single significant connection connecting the left postcentral gyrus-to-right postcentral gyrus. C, D Unsupervised hierarchical clustering results show univariate reduced data led to better group separation with the functional connectivity data. The figure shows the (C) alpha and (D) beta bands as examples, with the PTSD vs mTBI contrast. The dendrograms show the clustering for the participants (horizontal) and the univariate reduced features (vertical) based on the Z scores for AEC. The colour bar below the horizontal indicates the major (top three levels) participant clusters. The number of univariate reduced features are marked. Both C and D showed that, although with some exceptions, the univariate feature reduction helped reduce the functional connectivity data into features that separated the participants into PTSD and mTBI groups.