Digital Health, Journal Year: 2024, Volume and Issue: 10
Published: Jan. 1, 2024
Polysomnography (PSG) is unique in diagnosing sleep disorders, notably obstructive apnea (OSA). Despite its advantages, manual PSG data grading time-consuming and laborious. Thus, this research evaluated a deep learning-based automated scoring system for respiratory events sleep-disordered breathing patients.
Language: Английский
Computers in Biology and Medicine, Journal Year: 2023, Volume and Issue: 165, P. 107450 - 107450
Published: Sept. 9, 2023
Language: Английский
Citations
83
Neurocomputing, Journal Year: 2024, Volume and Issue: 577, P. 127317 - 127317
Published: Jan. 26, 2024
Language: Английский
Citations
58
Computers in Biology and Medicine, Journal Year: 2023, Volume and Issue: 160, P. 106998 - 106998
Published: May 6, 2023
Language: Английский
Citations
56
npj Digital Medicine, Journal Year: 2024, Volume and Issue: 7(1)
Published: June 1, 2024
Apnea and hypopnea are common sleep disorders characterized by the obstruction of airways. Polysomnography (PSG) is a study typically used to compute Apnea-Hypopnea Index (AHI), number times person has apnea or certain types per hour sleep, diagnose severity disorder. Early detection treatment can significantly reduce morbidity mortality. However, long-term PSG monitoring unfeasible as it costly uncomfortable for patients. To address these issues, we propose method, named DRIVEN, estimate AHI at home from wearable devices detect when apnea, hypopnea, periods wakefulness occur throughout night. The method therefore assist physicians in diagnosing apneas. Patients wear single sensor combination sensors that be easily measured home: abdominal movement, thoracic pulse oximetry. For example, using only two sensors, DRIVEN correctly classifies 72.4% all test patients into one four classes, with 99.3% either classified placed class away true one. This reasonable trade-off between model's performance patient's comfort. We use publicly available data three large studies total 14,370 recordings. consists deep convolutional neural networks light-gradient-boost machine classification. It implemented automatic estimation unsupervised systems, reducing costs healthcare systems improving patient care.
Language: Английский
Citations
7
Engineering Applications of Artificial Intelligence, Journal Year: 2024, Volume and Issue: 131, P. 107789 - 107789
Published: Jan. 1, 2024
Language: Английский
Citations
6
Operations Research Forum, Journal Year: 2023, Volume and Issue: 4(1)
Published: March 4, 2023
Abstract Understanding clinical features and risk factors associated with COVID-19 mortality is needed to early identify critically ill patients, initiate treatments prevent mortality. A retrospective study on patients referred a tertiary hospital in Iran between March November 2020 was conducted. COVID-19-related its association including headache, chest pain, symptoms computerized tomography (CT), hospitalization, time infection, history of neurological disorders, having single or multiple factors, fever, myalgia, dizziness, seizure, abdominal nausea, vomiting, diarrhoea anorexia were investigated. Based the investigation outcome, decision tree dimension reduction algorithms used aforementioned factors. Of 3008 (mean age 59.3 ± 18.7 years, 44% women) COVID-19, 373 died. There significant old age, low respiratory rate, oxygen saturation < 93%, need for mechanical ventilator, CT, cardiovascular diseases factor In contrast, there no gender, anorexia. Our results might help related better manage according extracted tree. The proposed ML models identified number patients. These if implemented setting needing medical attention care. However, more studies are confirm these findings.
Language: Английский
Citations
13
Sensors, Journal Year: 2023, Volume and Issue: 23(8), P. 3973 - 3973
Published: April 13, 2023
Sleep disorders can impact daily life, affecting physical, emotional, and cognitive well-being. Due to the time-consuming, highly obtrusive, expensive nature of using standard approaches such as polysomnography, it is great interest develop a noninvasive unobtrusive in-home sleep monitoring system that reliably accurately measure cardiorespiratory parameters while causing minimal discomfort user’s sleep. We developed low-cost Out Center Testing (OCST) with low complexity parameters. tested validated two force-sensitive resistor strip sensors under bed mattress covering thoracic abdominal regions. Twenty subjects were recruited, including 12 males 8 females. The ballistocardiogram signal was processed 4th smooth level discrete wavelet transform 2nd order Butterworth bandpass filter heart rate respiration rate, respectively. reached total error (concerning reference sensors) 3.24 beats per minute 2.32 rates for For females, errors 3.47 2.68, 2.33, verified reliability applicability system. It showed minor dependency on sleeping positions, one major cumbersome measurements. identified sensor region optimal configuration measurement. Although testing healthy regular patterns promising results, further investigation required bandwidth frequency validation larger groups subjects, patients.
Language: Английский
Citations
12
Deleted Journal, Journal Year: 2025, Volume and Issue: 48(2), P. 94 - 97
Published: Jan. 1, 2025
Citations
0
Biomedical Signal Processing and Control, Journal Year: 2025, Volume and Issue: 109, P. 107943 - 107943
Published: May 9, 2025
Language: Английский
Citations
0
Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)
Published: May 13, 2025
Adequate sleep is crucial for maintaining a healthy lifestyle, and its deficiency can lead to various sleep-related disorders. Identifying these disorders early essential effective treatment, which traditionally relies on polysomnogram (PSG) tests. However, diagnosing with high accuracy based solely electroencephalogram (EEG) signals, rather than using signals in complex PSG, reduce the time cost required, need specialized signal devices, as well increase accessibility usability. Previous studies have focused traditional machine learning (ML) methods such K-Nearest Neighbors (KNNs), Support Vector Machines (SVMs), ensemble analysis. models require manual feature extraction, prediction greatly depends type of extracted. Additionally, EEG datasets are small heterogeneous, challenging deep models. The study proposes an innovative multi-task convolutional neural network partially shared structure that uses frequency-time images generated from address limitations. proposed technique makes two predictions non-shared features time-frequency created through Short Time Fourier Transform (STFT) Continuous Wavelet (CWT), one features, final combination three predictions. weights this were optimized genetic algorithm Q-learning algorithm, aiming minimize loss maximize accuracy. utilizes dataset involving 26 participants examine impact Partial Sleep Deprivation (PSD) recordings. outcomes demonstrated model optimization methods, attained 98% test data predicting partial deprivation. This automated diagnostic efficient supporting tool rapidly effectively It swiftly precisely evaluates data, minimizing effort required by patient physician.
Language: Английский
Citations
0