Robotic rehabilitation for stroke patients has been an active field of research since the 1990s. There has been many studies focusing on mechanical design of robotic devices, design of software and interfaces for the patients and therapists, identifying quantitative and objective measures for motor improvement, and developing different operating modes/scenarios for the devices. However, a unified set of robotic (based on data captured by the robotic device) motor function improvement measures still does not exist. This makes comparing the outcomes obtained by using different robotic rehabilitation devices and platforms a difficult task. Typically, the motor improvement outcomes are reported using clinical stroke measures (such as Fugl-Meyer) that are reliable and widely used, albeit only in a pre-treatment and post-treatment fashion, which severely decreases the resolution of analyzing the actual recovery process.
We aimed to identify task and hardware independent robotic motor improvement measures that correlate well with clinical measures has remained an unaddressed need in the field. Defining such measures will make it possible to estimate clinical measures from a data file captured by the robotic device. Comparisons of the functional gains of patients that underwent different robotic rehabilitation protocols or used different devices will be more reliable and accurate when based on a unified set of robotic measures than when based on heterogeneous robotic measures or on pre- and post-treatment evaluations.
In our project in collaboration with The Institute for Rehabilitation and Reserach (TIRR), nine stroke patients underwent a hybrid robotic and traditional constraint-induced movement therapy (CIMT) for a month. During the robotic component of the rehabilitation protocol, patients completed a simple target-hitting task using a haptic joystick, which was capable of providing assistive/resistive forces or virtual fixture guides. Correlational analyses using several functional and impairment based clinical measures (Fugl-Meyer upper limb component, action research arm test, motor activity log, Jebsen-Taylor hand function test) and robotic measures (smoothness of movement, trajectory error and average number of target hits per minute) revealed strong and significant correlations exists between clinical and robotic mesaures. Initial results of the study with data from four patients were presented at IEEE ICRA 2008 (see publications below). The full study utilizing all of the mentioned clinical and robotic measures and data from all nine patients is reported in a journal manuscript that is currently under review.