ICA (Bell & Sejnowski, 1995) is a procedure to reconstruct a group of statistically independent components from a set of linearly mixed observations. Mathematically this can be written as X = AS, where X represents our observations, in this case regionally segmented brain volumes or multiple AUD assessments from each subject. S represents the latent, statistically independent components and each component in this study is a network from covaried brain regions or a factor from co-varied AUD assessments across subjects. The loading coefficients, A, represent how each of these components is represented in the subjects.
