We are currently extending the EEG feature extraction procedures by providing techniques for postextraction quality control. Although EEG is known for the relative high time investment required to produce clean, artifact‐free stretches of data, it is also quite unique for applying quality checks because of the large number of signals that are recorded from each subject. Each of these signals' extracted parameters can be matched against those of neighboring signals. Using spherical interpolation, signals can be recreated based on a fixed weighted average of all remaining electrodes, the EEG feature in question recalculated and matched against the original value (Junghöfer et al., 2000). Alternatively, machine learning can be used to establish an empirically estimated relation between the highly correlated values across the electrode locations and compare the actually obtained values to the values the model imputes from the data. Values with a deviance greater than expected may be removed or replaced by the imputed value. Figure 3 shows an example of how interpolation was used to detect rogue data points in the theta–beta ratio.
