EUVIMEDPoznaj EUVIMED
Open access · CC BY 4.0

A Multi-Frequency Self-Supervised Fusion Model for EEG-Based Dementia Classification

Ein Multi-Frequenz-Selbstüberwachtes Fusionsmodell für EEG-basierte Demenzklassifikation

Jinhua Sheng, Jiaqi Lin, Qiao Zhang, Zhaozhe Gong, Rong Zhang, Ming Wu, Yuhan Cheng, Xinrui Zhu, Zhongyuan Sun

Journal of Integrative Neuroscience · 2026-07-21 · 25 (7)

DOI: 10.31083/jin50737 ↗

Auf einen Blick

Das AM-SSF-Modell wurde zur Unterscheidung von Alzheimer-Krankheit und frontotemporaler Demenz aus EEG-Merkmalen über mehrere Frequenzbänder geprüft. Es erreichte 93,1 % Genauigkeit.

Fragestellung

Verbessert ein aufmerksamkeitsbasiertes Multi-Frequenz-Selbstüberwachtes Fusionsmodell die Klassifikation von Alzheimer-Krankheit und frontotemporaler Demenz aus EEG-Daten?

Methode

AM-SSF mit kontrastiven selbstüberwachten Encodern für θ-, α-, β- und γ-Band; aufmerksamkeitsgesteuerte adaptive Fusion; Random-Forest-Klassifikator; fünfzählige Kreuzvalidierung.

Ergebnisse

Das Modell erreichte 93,1 % Klassifikationsgenauigkeit in der fünfzähligen Kreuzvalidierung und schnitt besser ab als Single-Band-SSL und Average-Pooling-Fusion. θ- und β-Band trugen am meisten zur Entscheidung bei.

Schlussfolgerung der Autor:innen

Das AM-SSF-Modell verbessert die Klassifikation von AD und FTD und liefert Einblicke in die diskriminierende Rolle frequenzspezifischer Merkmale.

Kernaussagen

  • 93,1 % Genauigkeit in fünfzähliger Kreuzvalidierung
  • Besser als Single-Band-SSL und Average-Pooling-Fusion
  • θ- und β-Band hatten den größten Beitrag

Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.

At a glance

Journal article in Journal of Integrative Neuroscience, published online on 2026-07-21. Written by Jinhua Sheng and 8 co-authors. Institution: School of Computer Science, Hangzhou Dianzi University, 310018 Hangzhou, Zhejiang, China. Major MEDLINE topics: Alzheimer Disease, Electroencephalography, Frontotemporal Dementia, Supervised Machine Learning. Cited 1 times so far (Crossref); 31 references. The full text is freely available (open access, CC BY 4.0). Funded by: Key Program of Natural Science Foundation of Zhejiang Province, National Natural Science Foundation of China.

Auf einen BlickAbstraktSubjectsDane bibliograficzneCiteRelated studies

Abstract (original language)

Background:Brain source localization technology enables precise characterization of the spatial distribution of neural activity, serving as a crucial tool for exploring the pathological mechanisms underlying dementia. However, effectively integrating complementary diagnostic information from source localization features across multiple frequency bands remains a major challenge to enhancing classification performance and model interpretability.Methods:An attention-based multi-frequency self-supervised fusion model (AM-SSF) is proposed to address this issue. Independent contrastive self-supervised encoders are trained for the θ (4–8 Hz), α (8–13 Hz), β (13–30 Hz), and γ (30–48 Hz) frequency bands to learn band-specific latent representations. Then, an attention-guided adaptive fusion module is introduced to dynamically allocate band weights through cross-entropy–based supervised optimization, thereby achieving effective cross-band information integration. Finally, a random forest classifier is employed to evaluate the model’s performance in distinguishing Alzheimer’s disease (AD) from frontotemporal dementia (FTD).Results:Experimental results show that the proposed framework achieves a classification accuracy of 93.1% under five-fold cross-validation, significantly outperforming baseline methods such as single-band self-supervised learning (SSL) and average pooling fusion. Further analysis of the attention weight distributions revealed that the θ and β bands contributed most to model decision-making, providing interpretability regarding frequency-specific effects.Conclusions:In summary, the proposed AM-SSF model enhances AD and FTD classification performance while offering valuable insights into the discriminative roles of frequency band features.

Subjects

MeSH terms (MEDLINE)

Alzheimer DiseaseElectroencephalographyFrontotemporal DementiaSupervised Machine LearningBrain WavesHumansClassification Algorithms

Author keywords

Alzheimer’s diseaseattention mechanismbrain source localizationfrontotemporal dementiamulti-frequency fusionself-supervised learning

Authors and institutions

  • School of Computer Science, Hangzhou Dianzi University, 310018 Hangzhou, Zhejiang, China.
  • Key Laboratory of Intelligent Image Analysis for Sensory and Cognitive Health, Ministry of Industry and Information Technology of China, 310018 Hangzhou, Zhejiang, China.
  • Beijing Hospital, 100730 Beijing, China.
  • National Center of Gerontology, 100730 Beijing, China.
  • Institute of Geriatric Medicine, Chinese Academy of Medical Sciences, 100730 Beijing, China.

Funding

  • Key Program of Natural Science Foundation of Zhejiang Province
  • National Natural Science Foundation of China

Cite

APA

Sheng, J., Lin, J., Zhang, Q., Gong, Z., Zhang, R., Wu, M., Cheng, Y., Zhu, X., & Sun, Z. (2026). A Multi-Frequency Self-Supervised Fusion Model for EEG-Based Dementia Classification. Journal of Integrative Neuroscience, 25(7). https://doi.org/10.31083/jin50737

Vancouver

Sheng J, Lin J, Zhang Q, Gong Z, Zhang R, Wu M, et al. A Multi-Frequency Self-Supervised Fusion Model for EEG-Based Dementia Classification. Journal of Integrative Neuroscience. 2026;25(7). doi:10.31083/jin50737

BibTeX
@article{Sheng2026,
  author = {Sheng, Jinhua and Lin, Jiaqi and Zhang, Qiao and Gong, Zhaozhe and Zhang, Rong and Wu, Ming and Cheng, Yuhan and Zhu, Xinrui and Sun, Zhongyuan},
  title = {A Multi-Frequency Self-Supervised Fusion Model for EEG-Based Dementia Classification},
  journal = {Journal of Integrative Neuroscience},
  year = {2026},
  volume = {25},
  number = {7},
  doi = {10.31083/jin50737}
}
Start for free →