A Study on Changes in Estimation Accuracy for EEG Data During Calibration and Operation in MI-BCI

Changes in psychological factors have been suggested to cause variations in brain-computer interface (BCI) performance. More specifically, differences in psychological variables between the calibration and operation phases may cause a decrease in accuracy during operation, presenting a potential challenge for the adoption of BCI technology. The purpose of this study is to analyze the differences in accuracy between the calibration and operation phases of a BCI using a deep learning model. We structured tasks to simulate the calibration and operation phases, and participants performed motor imagery tasks under both conditions. The analysis revealed a significant decrease in accuracy for data obtained under the operation condition, highlighting the need for techniques capable of adapting to the electroencephalography signal data produced when users execute operations.