Test-Retest Reliability of Robotic Assessment Measures for the Evaluation of Upper Limb Recovery

• Published in: IEEE transactions on neural systems and rehabilitation engineering 2014-09, Vol.22 (5), p.1020-1029
• Main Authors: Colombo, Roberto, Cusmano, Ivana, Sterpi, Irma, Mazzone, Alessandra, Delconte, Carmen, Pisano, Fabrizio
• Format: Article
• Language: English
• Subjects: Adult; Aged; Algorithms; Biomechanical Phenomena; Female; Humans; Male; Middle Aged; Motor assessment; Movement; neurorehabilitation; Power measurement; Psychomotor Performance; Recovery of Function; Reliability; Reproducibility of Results; robot measures; Robot sensing systems; Robotics - methods; Robots; stroke; Stroke Rehabilitation; Systematics; Training; Treatment Outcome; Upper Extremity - physiopathology
• ISSN: 1534-4320; 1558-0210
• DOI: 10.1109/TNSRE.2014.2306571

Rehabilitation robots have built-in technology and sensors that allow accurate measurement of movement kinematics and kinetics, which can be used to derive measures related to upper limb performance and highlight changes in motor behavior due to rehabilitation. This study aimed to assess the test-retest reliability of some robot-measured parameters by analyzing their intra-session and inter-session (day-by-day) variability. The study was carried out in two groups: 31 patients after stroke and 15 healthy subjects. Both groups practiced two different motor tasks consisting of point-to-point reaching movements in the shape of two geometrical figures that were selected for the assessment of global and directional (eight directions of the workspace) test-retest reliability. The reliability of six parameters measuring movement velocity, accuracy, efficiency and smoothness was assessed intra-session and inter-session by the ICC, SEM, and CV. Healthy subjects exhibited very high ICC values ( > 0.85) and low SEM for all parameters. Patients had high ICC values and low SEM but their global reliability was generally lower compared to healthy subjects. In addition, their inter-session reliability showed very high ICC values ( > 0.91) and low SEM for all parameters. Direction analysis showed that in some parameters the reliability was generally high but not homogeneous in all directions. In addition, some directions showed systematic error. This study demonstrates that robot-measured parameters are reliable and can be considered ideal candidates for use in combination with impairment and functional clinical scales to evaluate motor improvement during robot-assisted neurorehabilitation.