An Extensive Investigation of Machine Learning Techniques for Sleep Apnea Screening

Abstract

The identification of Obstructive Sleep Apnea (OSA) relies on laborious and expensive polysomnography (PSG) exams. However, it is known that other factors, easier to measure, can be good indicators of OSA and its severity. In this work, we extensively investigate the use of Machine Learning techniques in the task of determining which factors are more revealing with respect to OSA. We ran extensive experiments over 1,042 patients from the Centre Hospitalier Universitaire of the city of Grenoble, France. The data included ordinary clinical information, and PSG results as a baseline. We employed data preparation techniques including cleaning of outliers, imputation of missing values, and synthetic data generation. Following, we performed an exhaustive attribute selection scheme to find the most representative features. We found that the prediction of OSA depends largely on variables related to age, body mass, and sleep habits more than the ones related to alcoholism, smoking, and depression. Next, we tested 60 regression/classification algorithms to predict the Apnea-Hypopnea Index (AHI), and the AHI-based severity of OSA. We achieved performances significantly superior to the state of the art both for AHI regression and classification. Our results can benefit the development of tools for the automatic screening of patients who should go through polysomnography and further treatments of OSA. Our methodology enables experimental reproducibility on similar OSA-detection problems, and more generally, on other problems with similar data models.

On Efficient Approximate Queries over Machine Learning Models

“…Example 2 (Preventive Medicine). Preventive medicine has many applications such as health promotion, early diagnosis, and personalized medicine [1,43]. One of the greatest obstacles of preventive medicine is the limited time a physician has [9,36].…”

“…1 ISSN 2150-8097. doi:XX.XX/XXX.XX image, given a distance measure [28,30]. Similarly, in the medical domain, a typical query would look for patients whose predicted clinical condition is similar to an input patient (see Figure 1) using a Deep Neural Network (DNN) [27,43]. A straightforward way of answering these queries is to apply the neural models exhaustively on all objects (here, images or patients) in the DB, and then return the objects that satisfy the query.…”

Developing Probabilistic Ensemble Machine Learning Models for Home-Based Sleep Apnea Screening using Overnight SpO2 Data at Varying Data Granularity

“…Numerous studies have leveraged a subset of signals derived from PSG, such as electroencephalogram (EEG), electrocardiogram (ECG), airflow, and blood oxygen saturation levels (SpO2), for automated sleep apnea screening. Performance outcomes vary based on signal modality and computational models utilized [4][5][6][7][8][9][10][11][12][13]. Yet, most sensing modalities, including EEG, ECG, and airflow, are not readily available for home use, hindering widespread adoption for at-home apnea screening.…”

Explainable Computer-Aided Detection of Obstructive Sleep Apnea and Depression

“…Keshavarz et al use the best-ranked features of 231 subjects' PSGs obtained via rank widget with information gain method with 10-fold crossvalidation as input to an ANN, NB, LR, KNN, SVM, and RF, of which ANN yields the best accuracy at 74.91 % [30]. Rodrigues Jr et al also aim to classify OSAS into one of 4 classes based on AHI, and in addition, aim to predict AHI with 60 classification and regression algorithms for data of the 1,042 subjects in the MARS dataset [31]. ExtraTrees yields the best results both as a regressor and classifier, as it yielded an R 2 value of 0.5, a root MSE (RMSE) of 16.25, a mean absolute error (MAE) of 9.6 for regression, and an accuracy of 75 %, an F1-score of 66 %, and an AUC of 85 % for classification [31].…”

“…Rodrigues Jr et al also aim to classify OSAS into one of 4 classes based on AHI, and in addition, aim to predict AHI with 60 classification and regression algorithms for data of the 1,042 subjects in the MARS dataset [31]. ExtraTrees yields the best results both as a regressor and classifier, as it yielded an R 2 value of 0.5, a root MSE (RMSE) of 16.25, a mean absolute error (MAE) of 9.6 for regression, and an accuracy of 75 %, an F1-score of 66 %, and an AUC of 85 % for classification [31]. Kim [33], [34].…”