This is the authors' abstract. We don't add key points for this paper.
Background Functional near-infrared spectroscopy (fNIRS) during cognitive tasks reflects coupled cortical, superficial/systemic and cardiorespiratory haemodynamics. We tested the signal basis of a reproducible cognitive task-related increase in output-only dynamic mode decomposition (DMD) mean frequency after upstream short-separation regression. Methods Total haemoglobin (HbT) from 95 prefrontal recordings in 76 adults was analysed over the first-to-last task-marker interval, split at its midpoint into INIT and LAST halves (≥120 s each): active healthy (n = 26), sedentary healthy (n = 22), and sedentary type 2 diabetes before exercise (n = 28), with 19 post-intervention recordings. The primary endpoint was the unweighted mean frequency of all returned DMD modes. Validation tested spectra, VLF-excluded common-component structure, rank/mode yield, preprocessing, pole composition, phase-surrogate nulls and filter-boundary controls. Results DMD frequency increased from 0.015 to 0.021 Hz at INIT to about 0.031 Hz at LAST (Hedges dz = 0.82–1.78). Full-band HbT first principal component (PC1) fraction decreased from 0.769 to 0.855 to 0.562–0.629. Removing 0.008–0.02 Hz eliminated approximately 89–91.5% of this decline while rank/mode-yield adjustment attenuated the ΔPC1–ΔDMD association. Within-segment surrogates bracketed the observed phase change, while surrogate mode-yield inflation tracked upward frequency bias (Spearman ρ = 0.743–0.848). Conclusions The DMD shift is closely associated with redistribution of shared very-low-frequency haemodynamic power and changes in dimensionality and pole composition.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Anirban Dutta (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: