Abstract
Obstructive sleep apnea is a heterogeneous sleep disorder with varying phenotypes. Several studies have already performed cluster analyses to discover various obstructive sleep apnea phenotypic clusters. However, the selection of the clustering method might affect the outputs. Consequently, it is unclear whether similar obstructive sleep apnea clusters can be reproduced using different clustering methods. In this study, we applied four well-known clustering methods: Agglomerative Hierarchical Clustering; K-means; Fuzzy C-means; and Gaussian Mixture Model to a population of 865 suspected obstructive sleep apnea patients. By creating five clusters with each method, we examined the effect of clustering methods on forming obstructive sleep apnea clusters and the differences in their physiological characteristics. We utilized a visualization technique to indicate the cluster formations, Cohen's kappa statistics to find the similarity and agreement between clustering methods, and performance evaluation to compare the clustering performance. As a result, two out of five clusters were distinctly different with all four methods, while three other clusters exhibited overlapping features across all methods. In terms of agreement, Fuzzy C-means and K-means had the strongest (kappa = 0.87), and Agglomerative hierarchical clustering and Gaussian Mixture Model had the weakest agreement (kappa = 0.51) between each other. The K-means showed the best clustering performance, followed by the Fuzzy C-means in most evaluation criteria. Moreover, Fuzzy C-means showed the greatest potential in handling overlapping clusters compared with other methods. In conclusion, we revealed a direct impact of clustering method selection on the formation and physiological characteristics of obstructive sleep apnea clusters. In addition, we highlighted the capability of soft clustering methods, particularly Fuzzy C-means, in the application of obstructive sleep apnea phenotyping.

Abstract

Good sleep is crucial for human life. Research has shown that poor-quality sleep is related to several cardiovascular and metabolic disorders. Sleep disorders are well categorized, and most of them have defined diagnostic criteria, with level 1 polysomnography being the gold standard. With the increasing use of technology, specifically smartphones, in people’s everyday lives, the search for alternative ways of monitoring sleep disorders or certain sleep parameters has been gaining relevance.

Abstract

Abstract
Obstructive sleep apnea (OSA) is a sleep disorder marked by repeated episodes of airway obstruction, leading to apneas (complete blockage) or hypopneas (partial blockage) during sleep. The standard diagnostic metric, the apnea-hypopnea index (AHI), quantifies the number of these events per hour of sleep but has limitations, such as its dependence on manual interpretation and lack of attention to event duration, which can be clinically significant. To address these issues, this study developed an algorithm to detect respiratory events from nasal airflow signals and measure their duration, using data from 22 patients at St. Vincent's University Hospital, sourced from the PhysioNet dataset. Signal processing techniques, including filtering and envelope analysis, were applied to extract features, and apnea/hypopnea events were identified based on American Academy of Sleep Medicine (AASM) guidelines. Events were classified by duration into three groups: 10-20 seconds, 20-40 seconds, and over 40 seconds. Preliminary results showed detection accuracy of 60% for apnea and 93% for hypopnea events. The study also explored relations between event duration and demographic factors, such as age, gender, body mass index (BMI), and Epworth Sleepiness Scale (ESS) scores, to assess whether longer events were linked to greater severity. These findings suggest that incorporating event duration and automated detection into OSA diagnosis could improve accuracy and provide better insight into the condition, potentially leading to more personalized treatments.

Abstract
This work introduces a web application for extracting, processing, and visualizing data from sleep studies reports. Using Optical Character Recognition (OCR) and Natural Language Processing (NLP), the pipeline extracts over 75 key data points from four types of sleep reports. The web application offers an intuitive interface to view individual reports' details and aggregate data from multiple reports. The pipeline demonstrated 100% accuracy in extracting targeted information from a test set of 40 reports, even in cases with missing data or formatting inconsistencies. The developed tool streamlines the analysis of OSA reports, reducing the need for technical expertise and enabling healthcare providers and researchers to utilize sleep study data efficiently. Future work aims to expand the dataset for more complex analyses and imputation techniques.

Abstract