Dementia severity index: A threshold-based approach to classifying dementia levels using resting state EEG

Area of Science:

• Neuroscience
• Biomarkers
• Medical Diagnostics

Background:

• Alzheimer's Disease (AD) and FrontoTemporal Dementia (FTD) present overlapping clinical symptoms, leading to frequent misdiagnoses.
• Current diagnostic methods like questionnaires and neuroimaging have limitations in subjectivity, cost, and accessibility.
• Electroencephalography (EEG) offers a cost-effective alternative, but quantitative analysis for dementia differentiation remains underexplored.

Purpose of the Study:

• To develop and validate a novel threshold-based approach for classifying dementia states using resting-state EEG.
• To introduce a Dementia Severity Index (DSI) as a quantitative EEG biomarker for differentiating AD, FTD, and Healthy Controls (HC).
• To explore EEG-based biomarkers capturing band-specific alterations associated with cognitive decline.

Main Methods:

• A threshold-based algorithm was developed for computing EEG biomarkers and formulating the Dementia Severity Index (DSI).
• The DSI was used to categorize individuals into AD, FTD, or HC groups.
• Classification performance was assessed using multiple machine learning classifiers and subject validation strategies across three EEG datasets.

Main Results:

• The DSI achieved classification accuracies up to 81.62% using kNN, demonstrating robust performance across diverse datasets.
• Threshold variation analysis confirmed the reliability of the approach.
• Significant correlations (0.79 and 0.62) were found between EEG-derived DSI predictions and actual Mini-Mental State Examination (MMSE) scores.

Conclusions:

• The proposed DSI provides an effective, threshold-based EEG framework for differentiating AD, FTD, and HC.
• This quantitative approach enhances interpretability, reduces subjective reliance in dementia assessment, and offers a potential clinical biomarker.
• The EEG-based method is minimally stressful for patients, making it suitable for clinical settings.