SSSI: A simulation-based framework for evaluating sigma-power fluctuation state identification pipelines in NREM sleep EEG

Sigma-band activity of the electroencephalogram during non-rapid eye movement (NREM) sleep exhibits infraslow fluctuations. These sigma-power fluctuations (SPFs) contain recurring local waveform periods, including ascending, high, descending, and low-sigma-power portions, which can be operationally categorized as SPF states. Although some SPF states are associated with arousal-related processes and sleep fragility, how the prevalence of SPF states changes within a continuous sleep episode remains unknown. Resolving this question requires a reliable analysis pipeline, yet it has been difficult to determine which pipeline best identifies SPF states. Here, we present the Simulation of SPF and SPF-State Identification (SSSI) framework, which generates simulated SPFs with known reference SPF states and uses them to evaluate candidate SPF-state identification pipelines. With a pipeline selected through SSSI, mouse NREM and human N2 sleep showed within-episode changes in SPF-state proportions in both species, with additional age-dependent variation in humans. Pipeline evaluation further showed that the detection of these effects can depend on the pipeline. Together, these findings establish SSSI as a framework for evaluating and interpreting within-episode temporal changes in SPFs while assessing pipeline-dependent methodological bias.