Method for predicting eye movement activity based on intelligent data analysis from a mobile portable electroencephalograph

Eye movement processes, such as fixations and saccades, play a crucial role in human cognitive activity as they are closely associated with functions such as perception, attention, and decision-making. These processes are actively applied in various fields, including human-computer interaction systems and neurophysiological research. Modern eye-tracking methods based on optical systems provide high accuracy but have several significant limitations. In this regard, the use of electroencephalographic (EEG) data for analyzing eye movement activity is becoming a promising direction, as EEG provides insights into the neural processes underlying eye movements. This allows many limitations of optical systems to be overcome, enabling the monitoring of eye movements without direct visual tracking. The goal of this study is to develop a method for predicting eye movement activity based on data collected from the BrainBit mobile EEG device. The study utilizes a long short-term memory (LSTM) neural network. Experimental results demonstrated that the proposed model achieves high accuracy in classifying the main types of eye movement activity. For fixations and saccades, the classification accuracy reached 91?%, indicating a high effectiveness of the model for these types of movements. However, classifying directed eye movements, such as left/right and up/down movements, proved to be a more challenging task, with an accuracy of about 65?% and 63?%, respectively. One of the key challenges identified in the study was the high individual variability of EEG signals between participants. As a result, the model training was personalized for each participant, which improved the accuracy of predictions for each individual. Thus, the study highlights the potential of using EEG for analyzing and predicting eye movements, particularly in the context of fixations and saccades.