Using resting-state fNIRS and graph-theory, the study found that Parkinson's patients with wearing-off exhibit a less efficient cortical network (reduced clustering and global efficiency, increased path length) with nodal abnormalities in premotor and somatosensory cortices, and that global…

Suggests a non-invasive, bedside-accessible candidate biomarker (reduced global efficiency) for objective detection and individualized monitoring of wearing-off, which could improve clinical management and outcome measures in PD trials.

Wearing-off (WO) is a common motor complication in Parkinson's disease (PD), characterized by the re-emergence of symptoms before the next dose of dopaminergic medication and still lacking objective, bedside-available neurophysiological biomarkers. In this study, we investigated cortical functional network alterations associated with WO using resting-state functional near-infrared spectroscopy (fNIRS), graph-theoretical analysis, and machine learning. Resting-state fNIRS signals were acquired over frontal and sensorimotor cortices, and functional connectivity matrices were used to derive global and regional network properties. Compared with PD patients without WO and healthy controls, patients with WO showed a less efficient cortical network topology, characterized by reduced clustering and network efficiency together with increased path length. Regional nodal abnormalities were most evident in premotor and somatosensory cortices. Classification analysis further supported the discriminative utility of these network features, with global efficiency emerging as a particularly informative metric. These findings suggest that resting-state fNIRS can detect clinically relevant cortical network alterations associated with WO and that reduced global efficiency may serve as a candidate non-invasive biomarker for objective identification and individualized monitoring of WO in PD.