Sleep Apnea Classification Using Deep Neural Network

Vattamthanam, Shyam; Mrudula, G B; Kumar, C. Santhosh

doi:10.1109/discover50404.2020.9278045

Cited by 7 publications

(1 citation statement)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We observed that the random forest, gradient boost ensembles and SVM based classifiers gave the best accuracy for the model trained with features generated from 10 and 30 minutes segmented signals. As reported in previous studies, the SVM and random forest classifier were found to be more apt for classification problems based on physiological signals [19]- [22].…”

Section: Discussionmentioning

confidence: 60%

Predicting Autonomic Dysfunction in Anxiety Disorder from ECG and Respiratory Signals Using Machine Learning Models

et al. 2021

View full text Add to dashboard Cite

<p class="0abstract">Anxiety is a cognitive, behavioural, and biological response that prepares the individual to handle the stresses and conflicts of everyday life. The excessive appearance of this biological response is diagnosed as an anxiety disorder, which is often associated with Autonomic dysfunction (ADy). ADy is difficult to study in clinics with very few parameters available. Detection of ADy may not be possible/difficult in anxiety disorder with the existing method. In this study, we built machine learning models to identify ADy in subjects with anxiety using properties extracted from ECG and respiratory signals. For each dataset, statistical and frequency domain features were estimated from ECG and respiratory signals. Supervised machine learning (ML) algorithms were used to classify the subjects. Out of 23 features estimated, 11 were found to be statistically significant for the classification. We segmented the signals into 5, 10, and 30 minutes intervals to build generalized models. To overcome data imbalance, ensemble techniques like boosting was used. The highest accuracy was obtained in the SVM, Random forest and Gradient Boosting classifiers (cross-validation accuracy of 82.2%, 81.64% and 79.06% and; AUC of 0.81, 0.76 and 0.84) for 10 and 30 minutes segmented datasets. Our results showed that the features extracted from the ECG signal are a good marker for diagnosing ADy in patients with anxiety disorder. Further, a deep neural network-based model can be implemented that may achieve better accuracy for classification provided with the cost of a large number of datasets and computation time.</p>

show abstract

Section: Discussionmentioning

confidence: 60%