Comprehensive Summary
Hou et al. confirmed the possibility of EEG microstate analysis as an Alzheimer's disease (AD) biomarker. EEG recordings of 36 AD patients and 29 normal patients were preprocessed, cropped to a length of 20 seconds, and classified into four characteristic microstates (A–D). Researchers calculated 24 microstate features (coverage, duration, frequency, and transition probabilities). Contrast statistical analysis revealed 21 differentiating characteristics between AD patients and normal patients. With these features, support vector machine classifiers were able to achieve a mean accuracy of 75.8% for AD patients and healthy participant discrimination. The study found that microstate abnormalities were linked to disruption of attention, cognitive control, and visual networks with implications of functional network disruption in AD. However, limitations such as low sample sizes and no testing across disease subtypes were also considered by the study.
Outcomes and Implications
This research is important as it shows EEG microstates as an inexpensive, accessible, and non-invasive screen for AD biomarkers. EEG is readily available compared to PET or CSF measurement and therefore best suited to detect at an early stage and screen very large numbers. Clinically, an EEG strategy will allow intervention to be initiated early, tracking disease progression, and even complement neuroimaging or fluid biomarkers. This being an early feasibility study of a dataset, the researchers state that this would require larger trials and external validation in most subtypes of AD before such an approach can be taken into practice.