The Cybathlon BCI race: Successful longitudinal mutual learning with two tetraplegic users

Abstract

This work aims at corroborating the importance and efficacy of mutual learning in motor imagery (MI) brain–computer interface (BCI) by leveraging the insights obtained through our participation in the BCI race of the Cybathlon event. We hypothesized that, contrary to the popular trend of focusing mostly on the machine learning aspects of MI BCI training, a comprehensive mutual learning methodology that reinstates the three learning pillars (at the machine, subject, and application level) as equally significant could lead to a BCI–user symbiotic system able to succeed in real-world scenarios such as the Cybathlon event. Two severely impaired participants with chronic spinal cord injury (SCI), were trained following our mutual learning approach to control their avatar in a virtual BCI race game. The competition outcomes substantiate the effectiveness of this type of training. Most importantly, the present study is one among very few to provide multifaceted evidence on the efficacy of subject learning during BCI training. Learning correlates could be derived at all levels of the interface—application, BCI output, and electroencephalography (EEG) neuroimaging—with two end-users, sufficiently longitudinal evaluation, and, importantly, under real-world and even adverse conditions.

Mindfulness Improves Brain Computer Interface Performance by Increasing Control over Neural Activity in the Alpha Band

“…Our work demonstrates, in a large sample, that important aspects of BCI training can be enhanced through a behavioral intervention (Mcfarland and Wolpaw 2018;Perdikis et al 2018). We identified the specific neural bases of the improvement in performance: Subjects trained in mindfulness were able to produce strikingly different patterns of alpha activity when the BCI task required volitional rest.…”

“…Whether the increased in alpha power over the motor cortex results from radiating alpha sources or traveling waves within or between disparate metastable brain states is as yet unknown (Vidaurre et al 2018;Zhang et al 2018;Roberts et al 2019). Regardless, since (1) most continuous BCI paradigms utilize rest as a state to be detected and (2) the decoder incorporates brain activity from both rest and motor imagery conditions in its decisions (i.e., there are not separate classifiers for rest and motor imagery), changing brain activity in one task improves classification of both conditions (evinced by the significant improvements in both 1D and 2D BCI control) (Perdikis et al 2018;Edelman et al 2019).…”

Spatiotemporal neural correlates of brain-computer interface learning

“…On the one hand, different approaches have been proposed to solve the problem by improving feature extraction and decoding algorithms (54), combining different modalities (65), or taking into account the user's profile (63). On the other hand, alternative accuracy metrics based on the separability of brain features, rather than on the simple count of successful control, have been shown to be more relevant for the proof-of-existence of subject's learning (66).…”

“…These findings suggest that BCI illiteracy could be an epiphenomenon biased by the nature of standard performance metrics which are affected by decoder recalibration (67), re-parameterizations of the BCI, and the application and adoption of better mental strategies (68; 59), among other factors. It is possible that other metrics, integrating both real performance and functional brain changes should be taken into account to better assess individual learning (66).…”

Augmentative and Alternative Communication (AAC) Advances: A Review of Configurations for Speech Disabled Individuals

“…where the variance including the user training, the associated ML, and the application in use [71,72]. Research in the area of BCIs is currently evolving [56], with promising results in recent state-of-the-art projects.…”