Predicting sleep quality among college students during COVID-19 lockdown using a LASSO-based neural network model

BMC Public Health, Journal Year: 2025, Volume and Issue: 25(1)

Published: Feb. 21, 2025

In March 2022, a new outbreak of COVID-19 emerged in Quanzhou, leading to the implementation strict lockdown management measures colleges. While existing research has indicated that pandemic had significant impact on sleep quality, specific effects containment college students' patterns remain understudied. This study aimed understand quality students Fujian Province during epidemic and determine sensitive variables, order develop an efficient prediction model for early screening problems students. A cross-sectional survey was conducted April 5-16, 2022 Quanzhou. total 4959 Quanzhou were enrolled this study. Descriptive analysis, univariate correlation multiple regression analysis used explore influencing factors regarding quality. addition, we constructed eight risk models predict mean PSQI score 6.03 ± 3.21 disorder rate 29.4% (PSQI > 7) obtained. Sleep latency, efficiency, diurnal dysfunction, all higher than national norm (P < 0.05). predictors finally identified by LASSO algorithm incorporated into models. Through series assessments, artificial neural network as best model, achieving area under curve 73.8% accuracy 67.3%, precision 84.0%, recall 66.3%, F1 69.3%. These performance indices suggest ANN outperforms other It is noteworthy threshold probabilities net benefit found be between 0.81 0.92 clinical confirmed models' predictions particularly effective identifying individuals with poor when probability set above 70%. findings underscore potential utility our detection disorders. quarantine management, affected certain extent, their scores average China. The performance, it expected provide interventions prevent

Language: Английский

---

Development of smart cardiovascular measurement system using feature selection and machine learning models for prediction of sleep deprivation, cold hands and feet, and Shanghuo syndrome

Measurement, Journal Year: 2023, Volume and Issue: 221, P. 113441 - 113441

Published: Aug. 11, 2023

This study aimed to develop a smart cardiovascular measurement system using ECG and PPG evaluate health issues: sleep deprivation, cold hands feet, the Shanghuo syndrome. The proposed methods extracted features from physical Signal utilized diverse machine learning techniques for evaluation. results demonstrated prediction accuracies exceeding 82% (87% deprivation k-nearest neighbor, 83% feet kernel classifier, syndrome ensemble learning). Moreover, this identified novel associated with in context of traditional Chinese medicine (TCM). An accurate TCM-defined syndrome, while considering relevant physiological features, is critical field research. developed can be seamlessly integrated existing instruments facilitate self-health management collaboration medical treatment.

Language: Английский

---

Potential of Machine Learning for Predicting Sleep Disorders: A Comprehensive Analysis of Regression and Classification Models

Diagnostics, Journal Year: 2023, Volume and Issue: 14(1), P. 27 - 27

Published: Dec. 22, 2023

Sleep disorder is a disease that can be categorized as both an emotional and physical problem. It imposes several difficulties problems, such distress during the day, sleep-wake disorders, anxiety, other problems. Hence, main objective of this research was to utilize strong capabilities machine learning in prediction sleep disorders. In specific, aimed meet three objectives. These objectives were identify best regression model, classification strategy highly suited datasets. Considering two related datasets evaluation metrics tasks classification, results revealed superiority MultilayerPerceptron, SMOreg, KStar models compared with twenty models. Furthermore, IBK, RandomForest, RandomizableFilteredClassifier showed superior performance belonged strategies. Finally, Function predictive among six considered strategies respect most metrics.

Language: Английский

---

Home Monitoring of Sleep Disturbances in Parkinson’s Disease: A Wearable Solution

2022 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops), Journal Year: 2024, Volume and Issue: unknown, P. 106 - 111

Published: March 11, 2024

Sleep Disorders are the most common and disabling non-motor manifestations of Parkinson's Disease (PD), significantly impairing quality life. Monitoring sleep disturbances in PD is a complex task, given lack objective metrics infrequent neurological assessments. This study proposes framework for detection patterns from data collected 40 subjects (12 PD) through wearable inertial measurement unit (IMU) during sleep, as well automatic assessment quality. Several features describing overnight motility proposed employed Machine Learning (ML) models to carry out classification. The best model achieved 96.2% Accuracy 93.4% F-1 score detecting controls, Leave-One-Subject-Out cross-validation approach. was assessed with an average accuracy 79.7% ± 4.4 across three tested classifiers, 75% 5.25 score. suggests feasibility characterising effectively monitoring symptoms' progression lightweight technology, pervasive, e-Health scenario.

Language: Английский

---

Knowledge, attitude, and practice toward sleep disorders and sleep hygiene among perimenopausal women

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: May 22, 2024

Abstract This cross-sectional study aimed to explore the knowledge, attitude and practice (KAP) toward sleep disorders hygiene among perimenopausal women, who were enrolled in Dezhou region of Shandong Province between July September 2023. A total 720 valid questionnaires collected (mean age: 51.28 ± 4.32 years old), 344 (47.78%) reported experiencing insomnia. The mean scores for attitude, practice, Dysfunctional Beliefs Attitudes about Sleep (DBAS) 15.73 7.60 (possible range: 0–36), 29.35 3.15 10–50), 28.54 4.03 6.79 1.90 0–10), respectively. Path analysis showed that knowledge had direct effects on ( β = 0.04, 95% CI 0.01–0.07, P 0.001), DBAS 0.02–0.05, < 0.001). Knowledge 0.11, 0.08–0.15, 0.001) indirect 0.02, 0.00–0.03, 0.002) effect practice. Moreover, also a impact 0.34, 0.25–0.43, In conclusion, women exhibited insufficient negative inactive hygiene, unfavorable DBAS, emphasizing need targeted healthcare interventions.

Language: Английский

---

A Personalized and Adaptive Distribution Classification of Actigraphy Segments into Sleep-Wake States

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Aug. 11, 2023

Abstract Wearable actimeters have the potential to greatly improve our understanding sleep in natural environments and long-term experiments. Current technologies served community well, but they known weaknesses that introduce errors can compromise reliable relevant clinical research wakefulness profiles from these data. Newer data collection technologies, such as microelectromechanical systems (MEMS), offer opportunities gather movement different forms at higher frequencies, making new analytical methods possible potentially advantageous. We developed a novel statistical algorithm, called Wasserstein Algorithm for Classifying Sleep Wakefulness (WACSAW), is based on optimal transport statistics uses MEMS its input. WACSAW segments group into periods with similar patterns generate profile each segment. The second utilization of methodology measures difference between segment hypothetical idealized sleep. Characteristic functions, derived individual activity segments, were clustered classified or wakefulness. was initially 6-person cohort applied an additional 16 independent participants. returned >95% overall accuracy assignments validated against participant logs. Compared Actiwatch Spectrum Plus, delivered ∼10% improvement accuracy, sensitivity, selectivity showed reduced standard error participants, indicating conformed individualized In addition, we directly compared GGIR, current used algorithm designed accept handled time series segmentation differently, which may contribute unique information recordings. Here, provide approach actimetry improves sleep/wakefulness designations, adapts individuals, provides interim metrics further interpretations open source modification. Author summary Wearables are emerging class real-time, important individuals their biological behavioral makeup. For 40 years, has been identifying living changes laboratory home environments. Yet, valuable analyses accuracy. wearables collect high frequency opportunity types analyses. (WACSAW). employs compare variation classify produces across 24-hr day both categorization more accurately classifies behavior than Plus. output be assess reliability requires no human intervention run. help achieve observations daily situations determine factors alter sleep, understand set stage diagnoses disease identification.

Language: Английский

---

Prediction of Sleep Health Status, Visualization and Analysis of Data

Proceedings of the International Conference on Advanced Technologies, Journal Year: 2023, Volume and Issue: unknown

Published: Aug. 19, 2023

Sleep, as an indispensable element of human life, is accepted one the main sources health, vitality and productivity. There are many factors that affect sleep health. Stress level, irregularity patterns excessive use technological devices can be given examples. Sleep health determined by analyzing various variables about sleep. using these with machine learning methods. For this purpose, a dataset containing 374 rows data 13 features was used in study. disorder conditions classified None, Apnea, Insomnia 12 features. Random Forest (RF), Support Vector Machine (SVM), Logistic Regression (LR) k Nearest Neighbor (kNN) methods were for classification. Classification success 91.66% from RF model, 90.27% SVM LR model 87.50% kNN model. In order to analyze which feature more effective classification processes, box plot correlation analysis used. As result analyzes, it body mass index has greatest effect on determination disorder.

Language: Английский

---

Potential of Machine Learning for Predicting Sleep Disorders: A Comprehensive Analysis of Regression and Classification Models

Published: Dec. 13, 2023

Sleep disorder is a disease that can be categorized as both an emotional and physical problem. It imposes several difficulties problems, such distress during the day, sleep-wake disorders, anxiety, other problems. Hence, main objective of this research to utilize strong capabilities machine learning in prediction sleep disorders. In specific, aims meet three objectives. These objectives are identify best regression model, classification strategy highly suits datasets. Considering two related datasets evaluation metrics tasks classification, results revealed superiority MultilayerPerceptron, SMOreg, KStar models compared with twenty-three models. Also, IBK, RandomForest, RandomizableFilteredClassifier showed superior performance belong strategies. Finally, Function predictive among six considered strategies respect most metrics.

Language: Английский

---

Detecting Anomalies of Daily Living of the Elderly Using Radar and Self-Comparison Method

Published: Nov. 28, 2022

Along with the aging society, elderly population increases. Most non-disabled prefer to age in their comfortable homes. To support such home care for elderly, continuous real-time monitoring of all this and early warning event an unexpected are beneficial. Current systems, as wearable sensors or webcams, could monitor activity people independent living. However, it malfunctions when do not wear sensors; webcam has privacy concerns. The study proposes a novel intelligent system daily life notify anomalies real time. Millimeter-wave (mmWave) radar, machine learning, self-comparison method were adopted implement system. A data-driven scheme is proposed reduce false alarms. Clinical data from 73 seniors (58 males; mean standard deviation 71.7 ± 7.4 years; 15 females; 70.8 7.8 years) collected hospital training sleep prediction model. Five older solidary volunteers attended collection at indoor tracking monitoring. experimental results revealed that achieve alarm rate below 5%. findings may serve guide research development non-invasive sensing systems adults home.

Language: Английский

---

A Hybrid and Ensemble Deep Learning Approach for Prediction and Analysis of Sleep Quality using Wearable IoT Device Data for Improved Accuracy

2022 International Conference on Inventive Computation Technologies (ICICT), Journal Year: 2023, Volume and Issue: unknown, P. 1737 - 1742

Published: April 26, 2023

Sleep quality refers to how well a person sleeps during the night. There are many factors that can affect sleep quality, including stress, anxiety, diet, exercise, and environmental such as noise light levels. Good is essential for overall of life. Poor have number detrimental impacts on one's physical mental health. To improve it important establish consistent routine. existing works prediction from wearable device data. Few those analyzed using same algorithms used in this study. Several machine learning algorithms, however, been proposed reach great accuracy. Overfitting insufficient data availability common problems these models. This research aims increase accuracy performance models predicting overcome challenges, objective work develop system combination feature selection techniques The methodology divided into three parts: preprocessing, model building, evaluation. Three types were study: single models, hybrid an ensemble training validation. acquired IoT was preprocessed by eliminating outliers normalizing trained evaluated based accuracy, precision, recall, F1-Score. results show superior all other terms F1-Score 0.9897 0.9745 respectively. had lower metrics compared model, but still performed better than individual provides insights potential devices demonstrates effectiveness combining different improved performance.

Language: Английский

---