Automated analysis of unattended portable oximetry by means of Bayesian neural networks to assist in the diagnosis of sleep apnea

Abstract

--Sleep apnea-hypopnea syndrome (SAHS) is a chronic sleep-related breathing disorder, which is currently considered a major health problem. In-lab nocturnal polysomnography (NPSG) is the gold standard diagnostic technique though it is complex and relatively unavailable. On the other hand, the analysis of blood oxygen saturation (SpO2) from nocturnal pulse oximetry (NPO) is a simple, noninvasive, highly available and effective alternative. This study focused on the design and assessment of a neural network (NN) aimed at detecting SAHS using information from at-home unsupervised portable SpO2 recordings. A Bayesian multilayer perceptron NN (MLP-NN) was proposed, fed with complementary oximetric features properly selected. A dataset composed of 320 unattended SpO2 recordings was analyzed (60% for training and 40% for validation). The proposed Bayesian MLP-NN achieved 94.2% sensitivity, 69.6% specificity, and 89.8% accuracy in the test set. Our results suggest that automated analysis of at-home portable NPO recordings by means of Bayesian MLP-NN could be an effective and highly available technique in the context of SAHS diagnosis.

Evaluation of Machine-Learning Approaches to Estimate Sleep Apnea Severity From At-Home Oximetry Recordings

“…Thus, we finally propose a comprehensive assessment of the athome SpO 2 usefulness by training and validating up to four new machine-learning derived models, ranging from simple to complex ones: linear discriminant analysis (LDA), logistic regression (LR), multi-layer perceptron Bayesian neural network (BY-MLP), and the ensemble learning method adaptive boosting (AdaBoost), arranged along with LDA as base classifiers (AB-LDA). Preliminary studies of our own group have been already conducted regarding BY-MLP and AdaBoost, providing signs of the usefulness of the machinelearning approach for at-home SpO 2 recordings [26], [27]. Nonetheless, the comprehensive approach conducted in the current study has led to the use of different SpO 2 information to train the models, as well as LDA instead of classification and regression trees as base classifiers for AdaBoost.…”

A Comprehensive Review: Computational Models for Obstructive Sleep Apnea Detection in Biomedical Applications

“…The symmetric uncertainty is calculated for each of the features by ranking using discriminant relevance (DR) method. Based on the rank obtained by them, the features are either selected or removed [ 36 ]. DR method is also used for feature ranking and selection.…”

“…The statistical analysis of the data is done using Kolmogorov-Smirnoff and Levene tests. The nonparametric Mann–Whitney U test is also done to identify the significant differences in the features [ 36 ]. The ANN along with the wavelet transforms is used for the detection of sleep apnea from EEG signals.…”

An Oximetry Based Wireless Device for Sleep Apnea Detection

“…The final point happens when the signal grows back to either 3% above the second point or 1% below the first point. The total time between the first and third points must be between ten and 90 s. A Feedforward Neural Network (FFNN) was employed by Álvarez et al [29] for OSA detection. The network was fed with features of the SpO2 signal, specifically: kurtosis, skewness, mean, relative power, spectral entropy, sample entropy, Lempel-Ziv complexity (LZC), and Central Tendency Measure (CTM).…”