Fig. 7: Longitudinal evaluation of PLE and LZC in a CLIS participant and a non-LIS participant from the ALS dataset (fluctuations between different sessions).

The mean PLE (a) and LZC (c) distributions, and their respective coefficients of variation (b, d) are depicted over sessions for the CLIS and non-LIS participants, as well as for the mean of the healthy controls group (as multisession records in controls are missing). An increase for mPLE (with a decrease for cvPLE) and a decrease for mLZC (with an increase for cvLZC) are observed for the CLIS participant as compared to the non-LIS participant and the average of the HC. Moreover, fluctuations between sessions in the CLIS patients were detectable in all measurements. PSD with standard error mean (e) showed a power spectrum close to the mean of HC for the patient with ALS without LIS. The PSD of the CLIS participant showed a generally steeper decay compared to HC and non-LIS, with higher power at slow frequencies (1–8 Hz), a slowdown and a shift in power in the delta-theta range with residual peak alpha activity, and a loss of power at higher frequencies (10–40 Hz). The correlation network (f) represents the dynamic variables of the brain through the nodes of the networks, while the significant Spearman rho correlations between them (p < 0.05) are represented by the blue links. Three measures derived from the EEG recordings (mean and CV of PLE and LZC) show a statistically significant correlation with the behavioral status assessed immediately after the recording. This association is lost in subsequent evaluations. Source data are provided in Supplementary Data 1.