Includes the code written for the "Viewer Engagement Analytics" project I undertook with Swapnil Chhabra and Ling Jin while at Northwestern University. Note that this code is not complete, rather the various components of our analysis were hacked together into this package so that those continuing the work could use it. All the ingredients required to replicate our analysis can be found here, and can be applied to any fMRI data provided in a similar format.
The aim of the project was to find whether there was a clear relationship between events viewed by subjects (e.g. in a movie) and the activity in their brains, as measured by blood oxygen levels from fMRI. In order to measure this activity, we looked at pairwise correlations between subjects over short time intervals, and summarised statistically significant average correlations across space (i.e. voxels) and time.
The analysis ultimately did not yield any clear findings, however it was a useful exercise in handling big data and understanding a little bit of neuroscience and how analytical techniques can contribute to the field.