Search

Stroke survivors with chronic paralysis often have difficulties to perform various activities of daily living (ADLs), such as preparing a meal or eating and drinking independently. Recently, it was shown that a brain/neural hand exoskeleton can restore hand and finger function, but many stroke survivors suffer from motor deficits affecting their whole upper limb. Therefore, novel hybrid electroencephalography/electrooculography (EEG/EOG)-based brain/neural control paradigms were developed for guiding a whole-arm exoskeleton. It was unclear, however, whether hemiplegic stroke survivors are able to reliably use such brain/neural-controlled device. Here, we tested feasibility, safety, and user-friendliness of EEG/EOG-based brain/neural robotic control across five hemiplegic stroke survivors engaging in a drinking task that consisted of several subtasks (e.g., reaching, grasping, manipulating, and drinking). Reliability was assumed when at least 75% of subtasks were initialized within 3 s. Fluent control was assumed if average “time to initialize” each subtask ranged below 3 s. System's safety and user-friendliness were rated using Likert-scales. All chronic stroke patients were able to operate the system reliably and fluently. No undesired side effects were reported. Four participants rated the system as very user-friendly. These results show that chronic stroke survivors are capable of using an EEG/EOG-controlled semiautonomous whole-arm exoskeleton restoring ADLs.

Este art́ıculo presenta un nuevo sistema de rehabilitación capaz de adaptarse al estado psicofisiológico del paciente durante tareas de rehabilitación robótica. Con este tipo de terapia se puede maximizar la motivación y participación del paciente durante la actividad de rehabilitación. En este trabajo se extienden los resultados del estudio presentado en (Badesa et al., 2014b), realizado con sujetos sanos, a su utilización con pacientes que hayan sufrido un accidente cerebrovascular. En una primera parte del art́ıculo se presentan los distintos componentes del sistema adaptativo, y se realiza una comparativa de distintas técnicas de aprendizaje automático para clasificar el estado psicofisiológico del paciente entre tres estados posibles: estresado, nivel de excitación media y relajado. Finalmente, se muestran los resultados del sistema autoadaptativo con un paciente con ictus en fase crónica, que modifica el comportamiento del robot de rehabilitación y de la tarea virtual en función de las medidas de las señales fisiológicas.