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

Abstract

AbstractBCI 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 μ 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 user’s ability to modulate μ 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 μ 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 user’s μ suppression during a MI-BCI task can be predicted based on their visuospatial memory ability, as measured by the Design Organization Test (DOT).