Schematic of the data acquisition setup (left panel) and main findings (right panel).

Each recording consisted of cortical activity (using electroencephalography) and respiratory activity (using impedance pneumography).

Cortico-respiratory coupling in infants.

A) Electroencephalographic (EEG) amplitude modulation depending on the respiratory cycle and frequency. EEG amplitude is averaged over breaths, recordings and channels FCz and Cz. The blue graph below is the impedance pneumographic signal time-locked from breath to breath and averaged over breaths and recordings. Grey and white bars below indicate expiration (Ex; in grey) and inspiration (In; in white). B) The time-resolved amplitude modulations averaged within four frequency bands (delta: 0.5-4 Hz, theta: 4-8 Hz, alpha: 8-13 Hz, beta: 13-25 Hz), focusing on channels FCz and Cz again. Shaded light-green area is the standard error across recordings after pooling over channels. C) Spatial map of coherence-based phase-amplitude coupling (PAC) between the respiratory phase (ωrespiratory phase) and EEG amplitude (ωEEG amplitude) – see Figure S1 for the pan-spectral PAC estimates. Spectra are the mean over all recordings. D) Spatial map for the corresponding statistical significance relative to surrogate PAC obtained through epoch shuffling (reported on a logarithmic scale).

Timing and directionality of the cortico-respiratory coupling.

A) Phase-amplitude coupling modulation (estimated as cross-frequency coherence) throughout the respiratory cycle. Graphs are colour-coded for the delta- and theta-frequency bands (dark- and light-blue colours respectively), which are the EEG’s primary frequency bands for which we observed statistically significant PAC (Figure 2C-D). Shaded regions indicate standard error over recordings. PAC was estimated for EEG channels FCz and Cz. Grey waves at the top and grey bar at the bottom indicate the phase of the respiratory cycle. Coloured dots indicate points that are statistically significant (alpha level was set to 0.001). B) Corresponding statistical significance analogous to the coloured dots at the top of panel A (black dashed line indicates the alpha level). C) Spatial map of the cross-frequency phase-slope index between respiratory phase (ωrespiratory phase) and EEG amplitude (ωEEG amplitude) (see Figure S2 for the pan-spectral estimates) and D) its statistical significance (reported on a logarithmic scale). The spectra show the mean over all recordings.

Cortico-respiratory coupling dependency on apnoea rate and postmenstrual age.

A) Relationship between apnoea rate and statistically significant phase-amplitude coupling (PAC; defined as coherence encompassing delta- and theta-band frequencies). Each blue dot is the data of an individual recording (data of 68 infants were included from 104 test occasions). Black linear graph is the best fit of the linear mixed-effects model (fixed factors: coupling averaged over statistically significant samples from channels FCz and Cz [see Figure S4 for channel-specific relationships], data length to determine apnoea rate, mode of ventilation, and postmenstrual age; random factor: infant). Panel title contains the test statistic of the predictor’s regression slope (t), its significance (p), and partial correlation coefficient (π). B-C) PAC spectra between the respiratory phase (ωrespiratory phase) and EEG amplitude (ωEEG amplitude) for B) preterm (58 recordings; 33.0-36.86 weeks), and C) very preterm (28 recordings; 28.0-32.86 weeks) infants. Both spectra are averaged over recordings and EEG channels FCz and Cz.

Infant demographics.

Reported values are mean and standard deviation or number of babies or recordings. Values are reported over recordings for ages (postmenstrual, gestational, and postnatal), respiratory distress syndrome, persistent pulmonary hypertension, infection and respiratory support. Birthweight, sex, mode of delivery, and Apgar scores are provided over infants.