Prediction of Inefficient BCI Users based on Cognitive Skills and Personality Traits

Abstract

BCI inefficiency is one of the major challenges of motor imagery brain-computer interfaces (MI-BCI). Past research suggests that certain cognitive skills and personality traits correlate with MI-BCI real-time performance. Other studies have examined sensorimotor rhythm changes (also known as mu suppression) as a valuable indicator of successful execution of the MI task. This research aims to combine these insights by investigating whether cognitive factors and personality traits can make predictions of a users ability to modulate mu rhythms during a MI-BCI task. Data containing 55 subjects who completed a MI task was employed, and a stepwise linear regression model was implemented to select the most relevant features for mu suppression prediction. The most accurate model was based on these factors: Spatial Ability, Visuospatial Memory, Autonomy, and Vividness of Visual Imagery. Further correlation analyses showed that a novice users mu suppression during a MI-BCI task can be predicted based on their visuospatial memory ability, as measured by the Design Organization Test (DOT).

Analyzing and computing humans by means of the brain using Brain-Computer Interfaces - understanding the user – previous evidence, self-relevance and the user’s self-concept as potential superordinate human factors of relevance

“…In this case, both past and current research suggest a strong correlation between the user's self-reported operation performance and their self-reported performance satisfaction with their BCI performance (see for a discussion Kübler et al, 2015). In addition, the user's self-reported and perceived effectiveness (Kübler et al, 2014) and their acceptability of the BCIs as well as their affinity with technology (e.g., Hagedorn et al, 2021;Leeuwis et al, 2021) seem to modulate BCI performance among different vulnerable end-user groups. This may include patients with stroke or with mental health conditions (e.g., Al-Taleb et al, 2019;Voinea et al, 2019).…”

Classification of motor imagery EEG using deep learning increases performance in inefficient BCI users

“…Traditionally, MI-BCIs operate on machine learning (ML) algorithms in which spatial features associated with movement imagination are recognized. The imagining of a left or right body movement is accompanied by a lateralization of event-related (de)synchronization (ERD/ERS) in the mu (7-13 Hz) and beta (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) frequency bands of EEG signals [7][8][9][10]. This brain activity feature is usually picked up by the Common Spatial Pattern (CSP) algorithm [11] and serves as an input to the ML algorithm classifying the imagined body movements.…”

“…This is called 'BCI illiteracy' [12] or 'BCI inefficiency' [13], where a user is considered unable to control a BCI, even after extensive training. While multiple studies have focused on identifying the inefficient users early on in research or adapting the BCI training to them [e.g., [14][15][16], the issue of BCI inefficiency might be argued more nuanced; as successful BCI control depends on a synergy between man and machine, and therefore enhancements on both sides are needed to reach efficient control [13].…”