Mark Sloan

Thank you for the feedback. I plan to stick with CDA.SCR (or the equivalent/perfectly correlated CDA.ISCR) for the final measure. Since ISCR is a measure of area (time integral), I decided to interpolate missing data points to achieve 960 samples, to keep identical lengths for the area measurement. To do this I calculated the overall slope (trendline) of the last 8 known measurements (at 128 Hz) and added the slope to new interpolated data points needed to get to 7,680 total samples. "8" was chosen...

I am comparing CDA.nSCR and CDA.SCR results from a series of data segments that each represent a 60-second experience within a VR video. The data segments range from 7,678 to 7,680 samples at 128 Hz (due to rare skips in the bluetooth GSR device). After downsampling to 16 Hz, the resulting data segments have either 959 or 960 samples. I have been direct comparing these, but when I truncate one sample to make all segments contain 959 samples, I sometimes see an increase of 1 CDA.nSCR counted peak,...

Thank you for your help. Yes, for some subjects the CDA.nSCR is similar between scenes (such as 10-16 range). Right now I plan to use CDA.SCR as the primary measuring stick for a 59-second response window [1 60].

Thank you for the response. I tried both [1 60] and [1 61] and got essentially the same result with a slightly smaller range for [1 61] and 1.000 correlation between them in Excel. Standardizing each window's CDA.SCR (by formula in Excel) produced almost identical results between them. I will plan to stick with the 59-second window [1 60]. Thank you for your help.

I am considering ways to account for the last second of SC data collected from a participant's 60-second VR experience. Currently, I identify one event at time '0' (the start of VR experience data) and a response window of [1 60], which produces a 59-second response window. I would like to extend to create a 60-second response window if possible [1 61]. Is this possible or do all calculations stop at the end of the available data (60 seconds)?

Attached are the results for SUBJECT-A.

Thank you for the response and advice. For SUBJECT-A I checked the scatter plots for nSCR and SCR for 13 scenes, and also checked for non-response and artifacts (results and downsampled data graphs attached). I did not notice any artifacts (downsampled 128 Hz to 16 Hz), and all data was within 2.7 to 3.9 MuS range, so not a non-responder. Could there be anything else I could check to assess why there is no correlation between CDA.nSCR and CDA.SCR? Thank you for your help and patience.

I have come up with a work around where I use the GUI to initially process the data and then further process with a MATLAB script into a consolidated .xls document for each subject. I use a combination of the "Save" .mat files (in the GUI) for log, tau, and error optimization information, and "era.mat" data from Event-Related Activation in the GUI results menu. This will take much longer than using the batch-mode, but at least I have a way forward that will work.