help > ROC curve analysis
Showing 1-2 of 2 posts
Nov 26, 2017 02:11 PM | Hua Yu
ROC curve analysis
Dear experts,
Recently, I made a two-sample t test to compare the clustering coefficient between the patients' and HCs' groups, and I found some group difference in two nodes , the DLPFC and vmPFC. Now I want to extract the value to do some simple ROC curve analysis to see if these group differences could clarify the two groups. Do you know what kind of value should I use?
Looking forward for your reply,
Thanks a lot,
Hua
Recently, I made a two-sample t test to compare the clustering coefficient between the patients' and HCs' groups, and I found some group difference in two nodes , the DLPFC and vmPFC. Now I want to extract the value to do some simple ROC curve analysis to see if these group differences could clarify the two groups. Do you know what kind of value should I use?
Looking forward for your reply,
Thanks a lot,
Hua
Nov 27, 2017 04:11 PM | Jeffrey Spielberg
RE: ROC curve analysis
You will want to extract the clustering coefficient values for
each ROI (for each participant). These are in the
X_propcalc.mat file (or the X_permout.mat file). You will
need to figure out the index # for the ROIs you want. To do
this, look in out.ROI_labels and find the lines corresponding to
the two ROIs. The line numbers will be the index values you
need. Note that this is not necessarily the same as the value
of that ROI in the original atlas you used. For example, if
the label for the dlPFC ROI was on line 30 of out.ROI_labels, the
values you want can be extracted by running:
dlPFC_CCvals = out.fullmat_graph_meas.clust_coef_pos(:,30,:);
This will give you an n x 1 vector containing the CC values for the dlPFC ROI (if you want the Zen-Horvath CC, use clust_coef_ZH_pos instead of clust_coef_pos).
dlPFC_CCvals = out.fullmat_graph_meas.clust_coef_pos(:,30,:);
This will give you an n x 1 vector containing the CC values for the dlPFC ROI (if you want the Zen-Horvath CC, use clust_coef_ZH_pos instead of clust_coef_pos).