500 Section - Kinematic Assessment with a 3-D Exoskeleton for Monitoring Upper Limb Motor Recovery in Subacute Stroke

Objectives : Upper limb robotic technologies provide quantitative metrics supporting neurorehabilitation in individuals with motor impairments post-stroke. While a few studies have explored planar 2-D end-effector robots, studies linking three-dimensional (3-D) robotic upper limb therapy outcomes to actual motor recovery remain rare. This observational study investigated whether kinematic parameters from a 3-D upper limb exoskeleton (Armeo Power, Hocoma) can effectively assess and monitor motor function in subacute stroke patients before (T0), during (TM), and after (T1) robot-assisted rehabilitation.

Design: Fifteen participants with moderate-to-severe hemiparesis underwent daily 40-min sessions with the 3-D Armeo Power exoskeleton for 25±3 sessions over five weeks, in addition to standard care. Clinical assessments (Fugl-Meyer Assessment Upper Limb [FMA-UL], Box and Block Test [BBT], Nine Hole Peg Test [NHPT], Frenchay Arm Test [FAT], Modified Ashworth Scale [mAS], Barthel Index [BI], modified Rankin Scale [mRS], A_FORCE, A_MOVE, A_GOAL) were performed at T0 and T1. Kinematic parameters (accuracy, efficiency, smoothness, speed, reaction time) were recorded for standard exercises (“Connect the Dots,” “Rainmug,” “Goalkeeper,” “Highflyer”) at T0, TM, and T1. Data were analyzed using repeated-measures ANOVA, and correlations with clinical outcomes via Pearson’s coefficient.

Results: Mean age was 58.7±13.3 years; 30% female; all right-handed; 50% right-sided hemiparesis; mean time from stroke onset 3.0±0.8 weeks. Significant improvements occurred between T0 and T1 in FMA-UL (67.0 [51–115] → 108.0 [84–119], p< 0.05), BBT (8.1±18.2 → 33.2±17.8, p< 0.05), and BI (49.0 [21–82] → 88.0 [85–100], p< 0.05). Shoulder ROM for abduction-adduction and rotation also improved (p< 0.05). Positive correlations were observed between FMA-UL and shoulder ROM (abduction-adduction: r=0.876, p=0.048; rotation: r=0.792, p=0.109). Kinematic analysis showed 85% of parameters differed significantly T0–T1 (p< 0.05), with accuracy and speed most sensitive.

Conclusion:

Kinematic data from a 3-D upper limb exoskeleton can effectively monitor neuromotor recovery post-stroke. Accuracy was the most sensitive metric for detecting functional improvements.