Sleep medicine · 2026
Predicting sleep state from continuous positive airway pressure flow in patients with obstructive sleep apnea.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Ahsan Muneeb, Chen Hsin-Yu, Chen Cheng-Yao J, Anwar Andira I, Tolbert Thomas M, Rapoport David M, Yaggi Henry K, Wu Hau-Tieng, Zinchuk Andrey V
Аннотация
INTRODUCTION: The first line treatment for obstructive sleep apnea (OSA) is continuous positive airway pressure (CPAP). CPAP adherence is assessed by hours of mask usage per night. To understand whether patients using CPAP are sleeping while wearing their mask (and thus may benefit from therapy), we aimed to predict sleep from CPAP flow signals.
METHODS: We developed a machine learning algorithm to predict sleep states (wake vs. sleep) using 130 features (126 derived from CPAP flow data, age, sex, body mass index (BMI) and CPAP level). We used polysomnography records from 100 randomly selected adult OSA patients undergoing CPAP titration at Yale Sleep Center (n = 70 training, n = 30 testing samples). Sleep stages were scored by AASM-certified technologists. We determined model accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).
RESULTS: The dataset included 48 women and 52 men with a median (interquartile) age of 61 (46.3-69.0) years, BMI of 33.8 (27.8-39.6) kg/m2 and residual apnea-hypopnea index of 3.0 (1.0-6.0) events/hour, with no significant differences between training and test samples. We used 69,520 (74% sleep) to train and 19,124 (76% sleep) epochs to test the model. The overall accuracy for sleep vs. wake was 0.82 with Cohen's Kappa of 0.51. The model demonstrated high sensitivity (86.4%) and PPV (89.0%) for sleep and lower sensitivity (66.4%) and PPV (60.7%) for wake.
CONCLUSION: Epoch-by-epoch sleep and wake can be effectively predicted by a model using CPAP flow and demographics. Such information may complement usage data from CPAP devices to better assess CPAP effectiveness.
METHODS: We developed a machine learning algorithm to predict sleep states (wake vs. sleep) using 130 features (126 derived from CPAP flow data, age, sex, body mass index (BMI) and CPAP level). We used polysomnography records from 100 randomly selected adult OSA patients undergoing CPAP titration at Yale Sleep Center (n = 70 training, n = 30 testing samples). Sleep stages were scored by AASM-certified technologists. We determined model accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).
RESULTS: The dataset included 48 women and 52 men with a median (interquartile) age of 61 (46.3-69.0) years, BMI of 33.8 (27.8-39.6) kg/m2 and residual apnea-hypopnea index of 3.0 (1.0-6.0) events/hour, with no significant differences between training and test samples. We used 69,520 (74% sleep) to train and 19,124 (76% sleep) epochs to test the model. The overall accuracy for sleep vs. wake was 0.82 with Cohen's Kappa of 0.51. The model demonstrated high sensitivity (86.4%) and PPV (89.0%) for sleep and lower sensitivity (66.4%) and PPV (60.7%) for wake.
CONCLUSION: Epoch-by-epoch sleep and wake can be effectively predicted by a model using CPAP flow and demographics. Such information may complement usage data from CPAP devices to better assess CPAP effectiveness.