Robust EEG Based Biomarkers to Detect Alzheimer's Disease.

Area of Science:

• Neuroscience
• Biomarker Discovery
• Medical Diagnostics

Background:

• Alzheimer's disease (AD) diagnosis requires accessible, low-cost methods.
• Current electroencephalogram (EEG) biomarkers lack robustness for clinical practice.
• There is a need for reliable EEG-based biomarkers for AD detection.

Purpose of the Study:

• To develop a methodological framework for robust EEG biomarker development for AD detection.
• To identify a combination of EEG biomarkers with clinically acceptable performance.
• To exploit the combined strengths of key EEG biomarkers for improved diagnostic accuracy.

Main Methods:

• Investigated a large number of existing and novel EEG biomarkers related to EEG slowing, complexity reduction, and connectivity decrease.
• Employed machine learning techniques, including Support Vector Machine and Linear Discriminant Analysis, for biomarker selection and model creation.
• Evaluated a total of 325,567 EEG biomarkers to identify an optimal panel.

Main Results:

• Identified a panel of six EEG biomarkers that demonstrated high diagnostic performance.
• Developed a diagnostic model with high sensitivity (≥85%) and perfect specificity (100%) for AD detection.
• Successfully combined multiple EEG biomarkers to overcome the limitations of single-marker approaches.

Conclusions:

• A panel of six EEG biomarkers can reliably detect Alzheimer's disease with high accuracy.
• This combined EEG biomarker approach offers a promising low-cost, easy-to-use method for AD diagnosis.
• The developed methodological framework can guide the creation of robust EEG biomarkers for clinical applications.