Phasic Intention Derivative Resilience and Multi-Lobe Cortical Reallocation During Prolonged Lower-Limb Exoskeleton Control: A 2D Source-Space Phase-Space Framework
- 1Independent Researcher, Tirunelveli, Tamil Nadu, India
Abstract
Non-invasive brain-computer interfaces (BCIs) for lower-limb exoskeleton control suffer from severe non-stationarity during prolonged ambulation, driven by central motor fatigue and electrode impedance drift. Standard classifiers relying solely on tonic primary motor cortex (M1) band power degrade rapidly as task duration extends. Here, we introduce a two-dimensional (2D) dynamical phase-space decoding framework, [W_P(t), dW_P(t)/dt], termed the Predictive Adaptive Temporal Tracking Unit (PATTU), which combines an autoregressive source-space intention state (W_P(t)) with its first-order phasic intention derivative (dW_P(t)/dt) reconstructed via anatomically constrained standardized low-resolution brain electromagnetic tomography (sLORETA). Evaluated across cross-session motor imagery (EEGBCI, N=10) and prolonged lower-limb exoskeleton ambulation (ds007788, N=7), adaptive phase-space boundary tracking restores cross-session classification accuracy from 55.87% to 65.51% (+9.64%, p = 0.0078) and achieves a cohort mean held-out ROC-AUC of 0.6730 during active exoskeleton control (peaking at 0.8257 in high-responder trials, with adaptive thresholding recovering accuracy from 53.76% to 73.76% in drift-affected runs). Longitudinal source-space tracking across 40-minute exoskeleton sessions reveals that while tonic paracentral M1 source power erodes under fatigue (declining in 5 of 7 subjects; -2.30% in sub-01), the phasic intention derivative remains resilient (1.202 s⁻¹ in early alert runs vs. 1.413 s⁻¹ in late fatigued runs; 119.5% mean retention), accompanied by a +150.0 ms relative within-subject onset latency shift (45.7 ms early vs. 195.7 ms late). Cortical surface mapping across six bilateral networks demonstrates that the central nervous system preserves phasic initiation velocity by dynamically reallocating relative source power from fatigued paracentral M1 generators toward prefrontal executive (+3.59%), supplementary motor, and primary somatosensory (S1, +1.34%) hubs.
Related
- From Professional Commitment to Continuity: How Resilience and Empowerment Reduce Turnover Intention Among Critical Care Nurses
- Proteomic Profiling of Medicago sativa Leaves upon Spodoptera litura (Lepidoptera: Noctuidae) Invasion: Activation of Defense Pathways and Photosynthetic Resource Reallocation
- Exoskeleton-based user's pose estimation: a hybrid geometric and machine learning approach
- Sex-related differences in whole body muscle activity during exoskeleton-assisted load carrying
- Investigating the role of phosphodiesterase Pde2 in coordinating resource reallocation through the environmental stress response
BibTeX
@misc{shajan2026phasic,
title = {{Phasic Intention Derivative Resilience and Multi-Lobe Cortical Reallocation During Prolonged Lower-Limb Exoskeleton Control: A 2D Source-Space Phase-Space Framework}},
author = {Nikil Shajan},
howpublished = {Zenodo},
year = {2026},
month = oct,
doi = {10.5281/zenodo.23479039},
url = {https://zenodo.org/records/23479039},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Phasic Intention Derivative Resilience and Multi-Lobe Cortical Reallocation During Prolonged Lower-Limb Exoskeleton Control: A 2D Source-Space Phase-Space Framework
AU - Nikil Shajan
PY - 2026
DA - 2026/10/19
PB - Zenodo
DO - 10.5281/zenodo.23479039
UR - https://zenodo.org/records/23479039
AB - Non-invasive brain-computer interfaces (BCIs) for lower-limb exoskeleton control suffer from severe non-stationarity during prolonged ambulation, driven by central motor fatigue and electrode impedance drift. Standard classifiers relying solely on tonic primary motor cortex (M1) band power degrade rapidly as task duration extends. Here, we introduce a two-dimensional (2D) dynamical phase-space decoding framework, [W_P(t), dW_P(t)/dt], termed the Predictive Adaptive Temporal Tracking Unit (PATTU), which combines an autoregressive source-space intention state (W_P(t)) with its first-order phasic intention derivative (dW_P(t)/dt) reconstructed via anatomically constrained standardized low-resolution brain electromagnetic tomography (sLORETA). Evaluated across cross-session motor imagery (EEGBCI, N=10) and prolonged lower-limb exoskeleton ambulation (ds007788, N=7), adaptive phase-space boundary tracking restores cross-session classification accuracy from 55.87% to 65.51% (+9.64%, p = 0.0078) and achieves a cohort mean held-out ROC-AUC of 0.6730 during active exoskeleton control (peaking at 0.8257 in high-responder trials, with adaptive thresholding recovering accuracy from 53.76% to 73.76% in drift-affected runs). Longitudinal source-space tracking across 40-minute exoskeleton sessions reveals that while tonic paracentral M1 source power erodes under fatigue (declining in 5 of 7 subjects; -2.30% in sub-01), the phasic intention derivative remains resilient (1.202 s⁻¹ in early alert runs vs. 1.413 s⁻¹ in late fatigued runs; 119.5% mean retention), accompanied by a +150.0 ms relative within-subject onset latency shift (45.7 ms early vs. 195.7 ms late). Cortical surface mapping across six bilateral networks demonstrates that the central nervous system preserves phasic initiation velocity by dynamically reallocating relative source power from fatigued paracentral M1 generators toward prefrontal executive (+3.59%), supplementary motor, and primary somatosensory (S1, +1.34%) hubs.
ER -