![](bg8.png)
nals depending on the status of obstructive apnea oc-
currence, in addition to depression status and sleep
stage. We then process the data to ensure it is clean,
has an approximately normal distribution, and is z-
score normalized before we partition and input it into
our three classifiers. We train three classifiers using
the intervals or observations of 75 % of the subjects
and perform 10-fold cross-validation on the same set,
then test classifier performance with the data of the
remaining 25 % of subjects. Using the Chi
2
algo-
rithm to select the six most important features and
ANN for classification yielded the best performance
with an accuracy of 79.00 %, F1-score of 80.00 %, a
κ of 0.58, a Matthews correlation coefficient of 0.58
and an AUC of 0.84, while also considering the low
computational cost compared to the GRU-LSTM. The
performance is promising, and we believe further pre-
processing of the data, as well as further optimizing
network architectures and hyperparameters and us-
ing more novel approaches like transformers could
improve classification performance. In addition, im-
plementing explainability metrics, like SHAP and de-
scriptions would certainly make our work more ac-
cessible to clinical personnel, or even laypersons.
ACKNOWLEDGEMENTS
The authors would like to thank the American Cen-
ter for Psychiatry and Neurology (ACPN) in Abu
Dhabi for their invaluable contribution in sharing the
polysomnography data and acknowledge the support
of the biomedical engineering department and the
Healthcare Engineering Innovation Center (HEIC) at
Khalifa University of Science and Technology. The
authors would also like to highlight the importance
of the KAU-KU Joint Research Program, in partic-
ular, project DENTAPNEA between Khalifa Univer-
sity and King Abdulaziz University, in particular the
advice of Dr. Angari, Dr. Balamesh, Dr. Khraibi, and
Dr. Marghalani.
REFERENCES
Al-Angari, H. (2008). Evaluation of chin emg activity at
sleep onset and termination in obstructive sleep apnea
syndrome. In 2008 Computers in Cardiology, pages
677–679. IEEE.
Bj
¨
ornsd
´
ottir, E., Benediktsd
´
ottir, B., Pack, A. I., Arnardot-
tir, E. S., Kuna, S. T., Gislason, T., Keenan, B. T.,
Maislin, G., and Sigurdsson, J. F. (2016). The preva-
lence of depression among untreated obstructive sleep
apnea patients using a standardized psychiatric inter-
view. Journal of Clinical Sleep Medicine, 12(1):105–
112.
Box, G. E. and Cox, D. R. (1964). An analysis of trans-
formations. Journal of the Royal Statistical Society:
Series B (Methodological), 26(2):211–243.
Cohen, J. (1960). A coefficient of agreement for nominal
scales. Educational and psychological measurement,
20(1):37–46.
Ejaz, S. M., Khawaja, I. S., Bhatia, S., and Hurwitz, T. D.
(2011). Obstructive sleep apnea and depression: a re-
view. Innovations in clinical neuroscience, 8(8):17.
Erdenebayar, U., Kim, Y. J., Park, J.-U., Joo, E. Y., and
Lee, K.-J. (2019). Deep learning approaches for au-
tomatic detection of sleep apnea events from an elec-
trocardiogram. Computer methods and programs in
biomedicine, 180:105001.
Fritsch, F. N. and Carlson, R. E. (1980). Monotone piece-
wise cubic interpolation. SIAM Journal on Numerical
Analysis, 17(2):238–246.
Gao, S., Calhoun, V. D., and Sui, J. (2018). Machine learn-
ing in major depression: From classification to treat-
ment outcome prediction. CNS neuroscience & thera-
peutics, 24(11):1037–1052.
Glorot, X. and Bengio, Y. (2010). Understanding the diffi-
culty of training deep feedforward neural networks. In
Proceedings of the thirteenth international conference
on artificial intelligence and statistics, pages 249–
256. JMLR Workshop and Conference Proceedings.
Hosseinifard, B., Moradi, M. H., and Rostami, R. (2013).
Classifying depression patients and normal subjects
using machine learning techniques and nonlinear fea-
tures from EEG signal. Computer methods and pro-
grams in biomedicine, 109(3):339–345.
Kahaner, D., Moler, C., and Nash, S. (1989). Numerical
methods and software. Prentice-Hall, Inc.
Khandoker, A. H., Palaniswami, M., and Karmakar, C. K.
(2008). Support vector machines for automated recog-
nition of obstructive sleep apnea syndrome from ECG
recordings. IEEE transactions on information tech-
nology in biomedicine, 13(1):37–48.
Matthews, B. W. (1975). Comparison of the predicted and
observed secondary structure of t4 phage lysozyme.
Biochimica et Biophysica Acta (BBA)-Protein Struc-
ture, 405(2):442–451.
Moradhasel, B., Sheikhani, A., Aloosh, O., and Dabanlou,
N. J. (2021). Chin electromyogram, an effectual and
useful biosignal for the diagnosis of obstructive sleep
apnea. Journal of Sleep Sciences, 6(1-2):32–40.
Moussa, M. M., Alzaabi, Y., and Khandoker, A. H. (2022).
Explainable computer-aided detection of obstructive
sleep apnea and depression. IEEE Access, 10:110916–
110933.
Mumtaz, W., Ali, S. S. A., Yasin, M. A. M., and Malik,
A. S. (2018). A machine learning framework involv-
ing EEG-based functional connectivity to diagnose
major depressive disorder (mdd). Medical & biologi-
cal engineering & computing, 56(2):233–246.
Murray, C. J., Vos, T., Lozano, R., Naghavi, M., Flax-
man, A. D., Michaud, C., Ezzati, M., Shibuya, K.,
BIOSIGNALS 2023 - 16th International Conference on Bio-inspired Systems and Signal Processing
210