Sleep EEG-Based Approach to Detect Mild Cognitive Impairment

Area of Science:

• Neuroscience
• Gerontology
• Biomedical Engineering

Background:

• Mild Cognitive Impairment (MCI) is an early precursor to Alzheimer's disease (AD), necessitating early detection for intervention.
• Existing methods primarily analyze awake electroencephalogram (EEG) recordings, potentially overlooking sleep-related neurophysiological changes.
• Emerging evidence links altered sleep architecture to cognitive decline, suggesting sleep EEG as a valuable diagnostic tool.

Purpose of the Study:

• To develop and validate a novel method for MCI detection using sleep EEG features.
• To investigate the utility of sleep EEG in characterizing neuroregulatory deficits associated with MCI.
• To compare the diagnostic performance of sleep EEG features against awake EEG features for MCI classification.

Main Methods:

• Analysis of sleep EEG recordings from 40 participants (20 MCI, 20 healthy controls).
• Extraction of specific sleep features, including slow waves and spindles, alongside spectral and complexity features.
• Classification of MCI using Support Vector Machine (SVM) and Gated Recurrent Unit (GRU) network models.

Main Results:

• The GRU network utilizing sleep EEG features achieved the highest MCI classification accuracy of 93.46%.
• Sleep EEG features demonstrated superior performance in distinguishing MCI from healthy controls compared to awake EEG.
• The study identified specific sleep EEG characteristics indicative of neuroregulatory deficits in MCI.

Conclusions:

• Sleep EEG analysis offers a promising, high-accuracy approach for early MCI detection.
• Incorporating sleep features into EEG analysis significantly enhances the ability to differentiate MCI from healthy cognition.
• This method holds potential for early intervention strategies to mitigate Alzheimer's disease progression.