File
Baldan F1, Turolla A1, Pregnolato G1, Agostini M1, Maistrello L1, Jakob I2, Longhi C1
1Laboratory of Neurorehabilitation Technologies, Fondazione Ospedale San Camillo IRCCS, Venice, Italy, 2Tyromotion GmbH, Graz, Austria
Background: Stroke is the first cause of disability worldwide and recovery of hand function represents the hardest target to achieve in neurorehabilitation. Robot-assisted therapy has been proved to be effective for upper limb recovery after stroke. Moreover, controlling rehabilitation device with surface EMG (sEMG) represents a valuable solution to amplify active stimulation of the central nervous system. Nevertheless, studies aimed at predicting recovery of hand function, are missing. Go beyond current limitations with the ability to deliver the best therapy to each individual patient according to the clinical picture, would allow to increase the number of patients undergoing early active therapy.
Purpose: The aim of this cross-sectional study was to identify which are the outcomes measures predictive for controlling a robot device by force or by sEMG, in a large cohort of stroke survivors experiencing impairment of hand function. The results of this study allowed to identify which patients enroll for EMG controlled robot-assisted rehabilitation.
Methods: We recruited 174 patients survived to an ischemic or hemorrhagic stroke and with both right or left lesion. Subjects were assessed with the following clinical scales: Fugl-Meyer Upper Extremity (F-M UE), Functional Independence Measure (FIM), Reaching Performance Scale (RPS), Box and Blocks Test (BBT), Modified Ashworth Scale (MAS), Nine Hole Pegboard Test (NHPT). Patients were asked to perform 5 minutes of hand opening/closing with the robot Amadeo® (Tyromotion GmbH, Graz, Austria) with sEMG control. The same task was than proposed with torque control. The ability to perform at least one hand opening/closing was considered as ability to control the device with none, only one or both tipe of control. ROC curves were calculated to test which of the outcome measures were the best predictors of the event.
Results: 60 patients weren´t able to control Amadeo® neither by sEMG or torque, 19 only by sEMG and 95 can actuate the robot both by sEMG and torque. The ability to perform sEMG control was determined by the following variables and the relative cutoff values: F-M UE motor section ≥ 24, F-M UE sensitivity section ≥ 23, Flexor Carpii spasticity ≤ 3 , MAS ≥ 4. The estimated probability to predict the control of the robot by sEMG according to these clinical features was 93%. With regard to the force, the probability to control the robot for patients with a F-M UE motor section higher than 24 points was estimated at 94%.
Conclusion(s): Results indicated that F-M UE was the best predictor of the ability to control by different modalities (i.e. sEMG, torque) a closed-loop robotic device for hand rehabilitation, with an accuracy higher than 90%.
Implications: This finding opens the possibility to plan personalized treatments based on patients' individual characteristics, thus allowing clinicians to plan specific therapies aimed at maximizing the highest outcome achievable after stroke.
Keywords: stroke, prognosis, rehabilitation
Funding acknowledgements: The research has been funded within the project "Myoelectric interfaces for motor control - MYOSENS" (FP7-EU-PEOPLE-2011-IAPP Grant agreement no. 286208)
Purpose: The aim of this cross-sectional study was to identify which are the outcomes measures predictive for controlling a robot device by force or by sEMG, in a large cohort of stroke survivors experiencing impairment of hand function. The results of this study allowed to identify which patients enroll for EMG controlled robot-assisted rehabilitation.
Methods: We recruited 174 patients survived to an ischemic or hemorrhagic stroke and with both right or left lesion. Subjects were assessed with the following clinical scales: Fugl-Meyer Upper Extremity (F-M UE), Functional Independence Measure (FIM), Reaching Performance Scale (RPS), Box and Blocks Test (BBT), Modified Ashworth Scale (MAS), Nine Hole Pegboard Test (NHPT). Patients were asked to perform 5 minutes of hand opening/closing with the robot Amadeo® (Tyromotion GmbH, Graz, Austria) with sEMG control. The same task was than proposed with torque control. The ability to perform at least one hand opening/closing was considered as ability to control the device with none, only one or both tipe of control. ROC curves were calculated to test which of the outcome measures were the best predictors of the event.
Results: 60 patients weren´t able to control Amadeo® neither by sEMG or torque, 19 only by sEMG and 95 can actuate the robot both by sEMG and torque. The ability to perform sEMG control was determined by the following variables and the relative cutoff values: F-M UE motor section ≥ 24, F-M UE sensitivity section ≥ 23, Flexor Carpii spasticity ≤ 3 , MAS ≥ 4. The estimated probability to predict the control of the robot by sEMG according to these clinical features was 93%. With regard to the force, the probability to control the robot for patients with a F-M UE motor section higher than 24 points was estimated at 94%.
Conclusion(s): Results indicated that F-M UE was the best predictor of the ability to control by different modalities (i.e. sEMG, torque) a closed-loop robotic device for hand rehabilitation, with an accuracy higher than 90%.
Implications: This finding opens the possibility to plan personalized treatments based on patients' individual characteristics, thus allowing clinicians to plan specific therapies aimed at maximizing the highest outcome achievable after stroke.
Keywords: stroke, prognosis, rehabilitation
Funding acknowledgements: The research has been funded within the project "Myoelectric interfaces for motor control - MYOSENS" (FP7-EU-PEOPLE-2011-IAPP Grant agreement no. 286208)
Topic: Robotics & technology
Ethics approval required: Yes
Institution: Fondazione Ospedale San Camillo IRCCS, Venezia, Italia
Ethics committee: Ethics Committee for Experimentation of ULSS12 and IRCCS San Camillo
Ethics number: Protocol 2014.14 – sERF
All authors, affiliations and abstracts have been published as submitted.
