Robust information transfer calculation reveals dynamic networks during sleep arousal

Figures

• 

  FIGURE 1. Distribution of sleep stages over the selected arousals. Non-rapid eye movement stages 1 and 2 have been combined to light sleep stage, and non-rapid eye movement stages 3 and 4 have been combined to deep sleep stage. REM, rapid eye movement.

• 

  FIGURE 2. Embedding for TE calculation from source X to destination Y. In this example, the history H (magenta) for source X contains five samples, and for destination Y, it contains three samples. The tau τ (orange) for X is one sample, and for Y, it is two samples. The delay d (red) is one sample.

• 

  FIGURE 3. Flowchart of the arousal selection, development, and analysis workflow, including method optimization and physiological investigation. Max. TE, maximum TE optimization; CV, coefficient of variation; GLMM, generalized linear mixed model.

• 

  FIGURE 4. Mean TE between investigated systems: parasympathetic nervous system (PNS), sympathetic nervous system (SNS), PNS + SNS, and central nervous system (CNS) across all sleep arousals during the respective sleep stage. REM: rapid eye movement. Statistical significant differences, p-values: *, p<0.05 ; **, p<0.01 .

• 

  FIGURE 5. Mean TE between investigated systems: PNS, SNS, PNS + SNS, and CNS for healthy controls and patients with cardiovascular diseases (CVD). Statistical significant differences: ***, p<0.001 .

• 

  FIGURE 6. Mean TE between investigated systems, PNS, SNS, PNS + SNS, and CNS.

• 

  FIGURE 7. Mean TE and 95% confidence interval over subject age (blue). Figure only included ages with more than 100 subjects (42–77 years). Linear trend of information transfer over subject age (red).

• 

  FIGURE 8. Mean TE of influence factors: (a) Sex, (b) apnea-hypopnea index (AHI) severity, and (c) arousal index severity. Statistical significant differences: *, p<0.05 ; **, p<0.01 ; ***, p<0.001 .