Fig. 2: Low-dimensional hierarchical index tracks fluctuations in multiple consciousness-related brain states.
From: Hierarchical fluctuation shapes a dynamic flow linked to states of consciousness
a The hierarchical index distinguished the sedation state from wakefulness/recovery at the individual level (**P < .01, wakefulness versus sedation: t = 6.96, unadjusted P = 6.6 × 10−7; recovery versus sedation: t = 3.19, unadjusted P = 0.0046; no significant difference was observed between wakefulness and recovery; two-sided paired t-test; n = 21 volunteers, each scanned in three conditions). b Top: distribution of the tendency of the hierarchical index to drift during a ~15 min resting-state scanning in HCP data (982 individuals × 4 runs; *P < 0.05, unadjusted, Pearson trend test); a negative correlation indicates a decreasing trend during the scanning; bottom: partial correlation (controlling for sex, age, and mean framewise distance) between the hierarchical index (averaged across four runs) and behavioral phenotypes. PC1 of reaction time and PSQI Component 3 were inverted for visualization (larger inter-individual hierarchical index corresponds to less reaction time and healthier sleep quality). c The hierarchical index captures the temporal variation in sleep stages in each of six volunteers (gray line: scores by expert; blue line: hierarchical index; Pearson correlation). The vertical axis represents four sleep stages (wakefulness = 0, N1 = −1, N2 = −2, slow-wave sleep = −3) with time is shown on the horizontal axis (Subject 2 and Subject 4 were recorded for 6000 s; the others summed up to 6750 s); For the visualization, we normalized the hierarchical indices across time and added the average value of the corresponding expert score. d Distribution of the hierarchical index in the Myconnectome project. Sessions on Thursdays are shown in red color (potentially high energic states, unfasting / caffeinated) and sessions on Tuesdays in blue (fasting/uncaffeinated). Applying 0.2 as the threshold corresponding to a classification accuracy over 80% (20 of 22 Tuesday sessions surpassed 0.2; 20 in 22 Thursday sessions were of below 0.2) e–f The hierarchical index can explain intra-individual variability in energy levels across different days (two-sided unadjusted Spearman correlation). The error band represents the 95% confidence interval. Source data are provided as a Source Data file.
