File
Z. Suica1, A. Kruse2,3, J. Taeymans2,4, C. Schuster-Amft1,5,6
1Reha Rheinfelden, Research Department, Rheinfelden, Switzerland, 2Bern University Applied Science, Department of Health Professions, Bern, Switzerland, 3Private Practice, Binningen, Switzerland, 4Vrije Universiteit Brussel, Brussels, Belgium, 5Bern University of Applied Sciences, Institute for Rehabilitation and Performance Technology, Burgdorf, Switzerland, 6University of Basel, Department of Sport, Exercise and Health, Basel, Switzerland
Background: Brain-computer interface (BCI) technology for rehabilitation of patients after stroke is rapidly developing. Several randomised controlled trials have investigated the effects of BCI technology on recovery of motor and brain function in patients after stroke.
Purpose: What is the effect of the application of BCI technology based on non-invasive EEG using motor imagery (MI) on motor and brain function recovery in patients after stroke?
Methods: A systematic literature search was performed in Medline, IEEE Xplore Digital Library, Cochrane, and Embase in July 2018 and was updated in March 2019. In total, 991 studies were identified and independently selected by two reviewers. Randomized controlled trials or controlled clinical trials that included BCI technology for improving motor and brain recovery in survivors of stroke were identified. Two independent reviewers assessed risk of bias using the Cochrane Collaboration tool to judge the risk of over- or underestimating the effects of the intervention. The GRADE approach was used to rate the overall quality of the evidence. The random-effects model of the Comprehensive Meta-Analysis software was used for data pooling and analysis. Standardized mean difference (SMD) with 95% confidence (95%CI) and prediction interval (95%PI) were calculated. Meta-regression was used to evaluate the effects of covariates on the pooled effect-size. Protocol registration with PROSPERO: CRD42018105832.
Results: Quantitative analysis showed that treatments in patients after stroke with BCI-assisted training compared to conventional therapy alone were effective with an SMD=0.39 (95%CI=0.17-0.62, 95%PI=0.13-0.66) for motor function recovery of the upper limb. Only moderate quality evidence exists that the effect remains after 12 weeks. For motor function recovery of the lower limb SMD=0.41 (95%CI=-0.29-1.12). BCI technology enhanced brain function recovery with SMD=1.11 (95%CI=0.64-1.59, 95%PI=0.33-1.89). Covariates (duration of training, level of impairment of upper limb, the combination of both) did not show significant effects on the overall pooled estimate.
Conclusion(s): This meta-analysis showed evidence that BCI training added to conventional therapy may enhance motor functioning of the upper extremity and brain function recovery in patients after a stroke. However, evidence for a positive effect of BCI training on motor functioning of the lower extremity in patients after stroke remains uncertain. Studies lacked a standardised evaluation of motor imagery ability of the included patients. Furthermore, the quality of the assessments evaluating neuropsychological aspects were difficult to judge due to lack of detail.
Implications: The assessments of neuropsychological parameters such as attention, concentration or cognition as well as MI ability measures, such as the Kinaesthetic and Visual Imagery Questionnaire (KVIQ) should be used to evaluate patients' capabilities important for the mental training. This would be an important aspect in the interdisciplinary collaboration of physiotherapists, occupational therapists and neuropsychologists in a context of a complex management of patients suffering from stroke.
Funding, acknowledgements: This study was not funded.
Keywords: Brain-Computer interface, stroke, randomized controlled trial
Topic: Neurology: stroke
Did this work require ethics approval? No
Institution: Reha Rheinfelden
Committee: Ethics Committee Northwest/Central Switzerland EKNZ
Reason: This SR is based on published data, and ethical approval is not required.
All authors, affiliations and abstracts have been published as submitted.
