Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Jul 17, 2026

DeepOmicsAE: Representing Signaling Modules in Alzheimer's Disease with Deep Learning Analysis of Proteomics, Metabolomics, and Clinical Data
09:47

DeepOmicsAE: Representing Signaling Modules in Alzheimer's Disease with Deep Learning Analysis of Proteomics, Metabolomics, and Clinical Data

Published on: December 15, 2023

Deep Model Families for EEG-Based Multi-Class Dementia Classification.

Soner Kotan1, Aydin Akan2

  • 1Turkish Health Data Research and Artificial Intelligence, Applications Institute Health Institutes of Turkey, Kadikoy 34718, Istanbul, Turkey.

International Journal of Neural Systems
|July 16, 2026
PubMed
Summary

Related Concept Videos

Dementia l: Introduction01:22

Dementia l: Introduction

Dementia is an acquired, progressive syndrome characterized by a decline in multiple cognitive domains severe enough to impair daily functioning and reduce independence. Although memory loss is a central feature, the diagnosis requires additional deficits involving language, executive function, visuospatial skills, judgment, calculation, or abstract reasoning. These cognitive impairments reflect underlying neurodegenerative or vascular processes that gradually disrupt neuronal networks...

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

On the Use of a Depth Camera for the Assessment of Upper Extremity Movements in Healthy Individuals.

Sensors (Basel, Switzerland)·2026
Same author

Null Subcarrier Index Modulation in OFDM Systems for 6G and Beyond.

Sensors (Basel, Switzerland)·2021
Same author

Epileptic seizure classifications using empirical mode decomposition and its derivative.

Biomedical engineering online·2020
Same author

Breast Cancer Detection with Reduced Feature Set.

Computational and mathematical methods in medicine·2015

Deep learning models show promise for dementia assessment using EEG data. Self-supervised learning, particularly SimCLR-Time, achieved the best performance in differentiating Alzheimer's disease, frontotemporal dementia, and healthy controls.

Area of Science:

  • Neuroscience
  • Artificial Intelligence
  • Biomedical Engineering

Background:

  • Deep learning (DL) shows potential for electroencephalography (EEG)-based dementia assessment.
  • Limited studies compare DL architectures using rigorous, leakage-free protocols for dementia classification.
  • Differentiating Alzheimer's disease (AD), frontotemporal dementia (FTD), and healthy controls (HC) using EEG requires robust evaluation.

Purpose of the Study:

  • To benchmark twelve DL architectures from four families (RNNs, TCNs, Transformers, SSL) for multi-class dementia differentiation using EEG.
  • To evaluate models under strict subject-independent cross-validation and subject-level aggregation for clinical relevance.
  • To assess classification performance, stability, efficiency, and interpretability of DL models for EEG-based dementia detection.
Keywords:
Alzheimer’s diseaseEEGdeep learningdementiafrontotemporal dementiainterpretabilityself-supervised learningtransformer

Related Experiment Videos

Last Updated: Jul 17, 2026

DeepOmicsAE: Representing Signaling Modules in Alzheimer's Disease with Deep Learning Analysis of Proteomics, Metabolomics, and Clinical Data
09:47

DeepOmicsAE: Representing Signaling Modules in Alzheimer's Disease with Deep Learning Analysis of Proteomics, Metabolomics, and Clinical Data

Published on: December 15, 2023

Main Methods:

  • Benchmarking twelve DL architectures (RNNs, TCNs, Transformers, SSL) on a clinical EEG dataset (36 AD, 23 FTD, 29 HC).
  • Utilized subject-independent five-fold cross-validation to prevent data leakage.
  • Aggregated segment-level predictions to the subject level using majority voting for clinical relevance.

Main Results:

  • The self-supervised SimCLR-Time framework achieved the highest subject-level performance (mean accuracy [Formula: see text], weighted F1-score [Formula: see text]).
  • The Time-Series Transformer (TST) was the top fully supervised model (accuracy [Formula: see text]), followed by LSTM-FCN ([Formula: see text]).
  • TCN-Attention offered a good accuracy-efficiency trade-off, while SimCLR-Time demonstrated fast inference. Interpretability analysis indicated reliance on spatiotemporal patterns.

Conclusions:

  • Self-supervised and attention-based DL architectures show significant promise for EEG-based dementia decision-support systems.
  • This study provides a controlled benchmark for multi-class dementia classification from raw EEG.
  • External validation on larger, diverse cohorts is necessary for broader generalization of findings.