Test-retest reliability of kinematic and EEG low-beta spectral features in a robot-based arm movement task

Abstract Objective. Low-beta (L β , 13–20 Hz) power plays a key role in upper-limb motor control and afferent processing, making it a strong candidate for a neurophysiological biomarker. We investigate the test-retest reliability of L β power and kinematic features from a robotic task over extended intervals between sessions to assess its potential for tracking longitudinal changes in sensorimotor function. Approach. We designed and optimized a testing protocol to evaluate L β power and kinematic features (maximal and mean speed, reaction time, and movement duration) in ten right-handed healthy individuals that performed a planar center-out task using a robotic device and EEG for data collection. The task was performed with both hands, and the experiment was repeated approximately 40 days later under similar conditions, to resemble real-life intervention periods. We first characterized the selected features within the task context for each session, then assessed intersession agreement, the test-retest reliability (Intraclass Correlation Coefficient, ICC), and established threshold values for meaningful changes in L β power using Bland-Altman plots and repeatability coefficients. Main Results. L β power showed the expected contralateral reduction during movement preparation and onset. Both L β power and kinematic features exhibited good to excellent test-retest reliability (ICC > 0.8), displaying no significant intersession differences. Kinematic results align with prior literature, reinforcing the robustness of these measures in tracking motor performance over time. Changes in L β power between sessions exceeding 11.4% for right-arm and 16.5% for left-arm movements reflect meaningful intersession differences. Significance. This study provides evidence that L β power remains stable over extended intersession intervals comparable to rehabilitation timelines. The strong reliability of both L β power and kinematic features supports their use in monitoring upper-extremity sensorimotor function longitudinally, with L β power emerging as a promising biomarker for tracking therapeutic outcomes, postulating it as a reliable feature for long-term applications.