EUVIMEDEUVIMED entdecken
Crossref

A Comparative Performance Analysis of CNN, LSTM, and CNNLSTM Models for Classifying Sleep Stages Using Electroencephalography Signals

Zainab N. Nemer, Saba Abdual Wahid Saddam, Rana J. AL-Sukeinee, Esra'a J. Harfash

Informatica · 21.07.2025 · 49 (28)

DOI: 10.31449/inf.v49i28.8082 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Informatica, online erschienen am 21.07.2025. Verfasst von Zainab N. Nemer und 3 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper aims to study and analyze the performance of CNN, LSTM and CNN+LSTM deep neural networks individually in the problem of sleep cycle stage classification, by analyzing the recorded EEG signals, and highlight the strengths of each model in this classification problem. The deep learning models selected in this work are reliable models in the world of experiments and medical classification. The data used here is the EEG signal represented by the sleep-edf-expanded-npzs database which carries information about the different sleep stages, which are five stages: wakefulness (W), non-REM stage 1 (N1), non-REM stage 2 (N2), non-REM stage 3 (N3), and rapid eye movement (REM). This data was split by training by 70% and testing by 30%. The models were trained based on cross-entropy loss and Adam optimizer. The results obtained in this study showed that the classification rates are as follows: CNN model accuracy (0.8736%), LSTM model accuracy (0.8306%), and LSTM + CNN hybrid model accuracy (0.8813%). These results indicate the good performance of the CNN model, which can be attributed to its strong ability to extract spatial features from data. The results of the LSTM approach demonstrate its ability to track and interpret the temporal characteristics of the sleep signal. For the CNN-LSTM model, which combines the strengths of temporal LSTM and spatial CNN, the result of combining features in this sequence showed superior results than either CNN or LSTM alone in processing sleep signals. Statistical measures were used to validate the results obtained. The F-statistic was 949.78666666523 and the probability value was 6.119863 e-14, which validate the results obtained for each model.

Zitieren

APA

Nemer, Z. N., Saddam, S. A. W., AL-Sukeinee, R. J., & Harfash, E. J. (2025). A Comparative Performance Analysis of CNN, LSTM, and CNNLSTM Models for Classifying Sleep Stages Using Electroencephalography Signals. Informatica, 49(28). https://doi.org/10.31449/inf.v49i28.8082

Vancouver

Nemer ZN, Saddam SAW, AL-Sukeinee RJ, Harfash EJ. A Comparative Performance Analysis of CNN, LSTM, and CNNLSTM Models for Classifying Sleep Stages Using Electroencephalography Signals. Informatica. 2025;49(28). doi:10.31449/inf.v49i28.8082

BibTeX
@article{Nemer2025,
  author = {Nemer, Zainab N. and Saddam, Saba Abdual Wahid and AL-Sukeinee, Rana J. and Harfash, Esra'a J.},
  title = {A Comparative Performance Analysis of CNN, LSTM, and CNNLSTM Models for Classifying Sleep Stages Using Electroencephalography Signals},
  journal = {Informatica},
  year = {2025},
  volume = {49},
  number = {28},
  doi = {10.31449/inf.v49i28.8082}
}
Kostenlos starten →