<p>Eye-based interaction systems have significantly enhanced the quality of life for individuals with disabilities by restoring their communication abilities. These systems require users to fixate their gaze on a key for a specific duration until it is fully entered. In this study, we aim to evaluate the correlation between the user’s mental attention and gaze focus during key selection, and explore whether this potential connection can be harnessed to improve the efficiency of eye-based interaction systems. To address this issue, we have proposed a method for detecting user attention, while using an eye-controlled on-screen keyboard by integrating EEG and Eye-Tracking (ET) signals. Our hypothesis posits that cognitive attention and gaze coincide during key selection. After data collection, EEG signals are labeled based on ET signals, followed by several preprocessing steps. Differential entropy and latent vectors via Variational Autoencoder are used as EEG features, while fixation and saccade serve as ET features. A Convolutional Neural Network is employed to combine these features to determine the user’s level of attention. We recorded the EEG and ET signals of 30 healthy subjects using a Vietnamese eye-controlled spelling communication system. Our method achieves a classification accuracy of up to 92.37% with k-fold cross-validation and 96.80% with cross-subject validation. These findings indicate a correspondence between the user’s cognitive attention and gaze during key selection. The proposed method can enhance the efficiency of eye-based interaction systems by improving key selection speed and developing systems to monitor and alert individuals when they lose focus during activities such as driving or learning.</p>