Figure 3

Validation using the wake only and the sleep data sets. (A) Probabilities of observing wakefulness (P(W)) and any NREM sleep stage (P(S)) in the data set containing sleep (“sleep data set”) based on the EEG scoring (left column, “EEG sleep”) and the matched clustered dynamic connectivity states (middle column, “Clustering sleep”) and respective probabilities for the clustered dynamic connectivity states in the wake only data set (right column, “Clustering wake”). For the latter data set, probabilities for the EEG-based scoring are not shown (P(W) equaling 1 and P(S) equaling zero by definition). All plots are mean ± SEM. (B) Distribution of accuracies obtained using a bootstrapping procedure with 10 randomly selected participants and 100 iterations, shown for the data set containing sleep epochs (red, “Sleep”), the wake only data set (blue, “Wake”) and the sleep data set with randomized sleep stage labels (black, “Random”).