Precise Discrimination for Multiple Etiologies of Dementia Cases Based on Deep Learning with Electroencephalography

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Diagnostics

Background:

• The global rise in dementia necessitates accessible diagnostic biomarkers.
• Electroencephalography (EEG) offers a sensitive, inexpensive, and widely available method for dementia screening.
• Deep learning algorithms enhance EEG analysis for accurate, automated dementia classification in clinical settings.

Purpose of the Study:

• To evaluate the diagnostic accuracy of a novel deep learning algorithm using EEG data for differentiating dementia patients from healthy individuals.
• To assess the algorithm's ability to distinguish between Alzheimer's disease (AD), dementia with Lewy bodies (DLB), and idiopathic normal pressure hydrocephalus (iNPH).

Main Methods:

• A novel deep neural network was employed to analyze EEG data.
• EEG recordings from 55 healthy volunteers (HVs) and 301 patients (101 AD, 75 DLB, 60 iNPH) were analyzed.
• The network's discriminative accuracy for various dementia types was evaluated.

Main Results:

• The deep learning algorithm achieved high discrimination rates between HVs and dementia patients.
• Specific accuracies included 81.7% vs. AD, 93.9% vs. DLB, and 93.1% vs. iNPH.
• Discrimination against a combined group of AD, DLB, and iNPH was 87.7%.

Conclusions:

• EEG data analyzed by a novel deep learning algorithm successfully discriminated dementia patients from healthy individuals.
• This method shows potential for automated screening and diagnostic assistance for dementia diseases.
• The findings support the integration of AI-powered EEG analysis in clinical dementia diagnostics.