Fig. 4

Investigating modules of transitions between whole-brain network states. a The figure shows the 19 × 19 transition probability matrix of the HMM states calculated for the 18 participants that included all four PSG stages in their respective scanning session. This quantifies the likelihood of transitioning from any given state to any other state, yielding each matrix entry: transition probability from departure state to destination state. b A few HMM states were ‘sporadic’ and did not occur consistently across participants. HMM states not occurring in more than 25% of the participants were excluded. c The strongest transitions of the consistent HMM states were partitioned through a modularity analysis, and reorganised in a matrix according to the four resulting modules. d The transitions shown in c are presented as a transition map with each state depicted as a pie plot expressing its specificity for each of the four PSG stages. Arrows show the direction of the transitions with thickness proportional to the transition probability. The transitions describe a passage from HMM states with high activity during wakefulness in the top, further down through HMM states including more N1, and down to HMM states specific to N2 and finally N3. Interestingly, wakefulness appears to be represented by two modules (red and black). Even though no individual HMM state showed clear specificity for N1 sleep, a module (white) is evident between HMM states with specificity for wakefulness and HMM states with specificity for N2 and N3 sleep. e Pie chart showing the total proportion PSG stages within the 18 participants. Transition prob.: transition probability