Applying the Eye-Tracking Method for the Classification of Neurological Disorders, Mental Diseases, and Speech Impairments Based on Machine Learning: an Overviewстатья
Информация о цитировании статьи получена из
Scopus
Дата последнего поиска статьи во внешних источниках: 15 февраля 2024 г.
Аннотация:Eye tracking is a non-invasive technology that
facilitates the real-time monitoring of eye movements. It offers
insights into visual behavior and cognitive processes. The analysis
of eye movement data involves two key components: fixations,
(periods of stable gaze), and saccades (rapid shifts in focus between
points of interest). This brief overview evaluates significant recent
studies conducted between 2018 and 2023 that employ eye
tracking for task related to the medical classification. Main
algorithms that process eye movement data are used to extract
features like pupil positions, fixation duration, and saccade
characteristics. Our analysis showed that investigators use
machine learning algorithms such as support vector machines, knearest
neighbors, random forest, and convolutional neural
networks for distinguishing normal and pathological states,
typically in binary classification tasks, measured by accuracy and
AUC. Generally, the size of the datasets is limited. However,
authors achieved reliable classification results, ranging from 52%
to 95%. As technology continues to evolve, the integration of eye
tracking and machine learning offers a promising path toward
enhancing our understanding of cognitive processes and medical
diagnostic capabilities.