New MCI Detection Method Based on Transformer and EEG Data

Preventing health issues is a crucial aspect of the medical field, particularly when it comes to mild cognitive impairment (MCI), which is a risk factor for developing dementia. Early detection of MCI is essential, and there are two types of MCI: amnestic MCI (aMCI) and non-amnestic MCI (naMCI). However, it is challenging to differentiate between individuals with MCI and those who are aging normally. Electroencephalography (EEG) is a promising modality for diagnosing MCI, which provides information about an individual's cognitive state during a clinical examination. This research aims to distinguish between individuals diagnosed with MCI as either aMCI or naMCI, and healthy controls (HC) during a verbal fluency task (VFT). To achieve this, a new MCI detection method based on the transformer architecture was proposed. This method makes use of EEG data and achieves up to 94.78% accuracy.