Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Statistical learning for the analysis of multimodal sleep in older men

Zhu, Jianhui (2019) Statistical learning for the analysis of multimodal sleep in older men. Master's Thesis, University of Pittsburgh. (Unpublished)

Submitted Version

Download (1MB) | Preview


Introduction: Sleep is essential for human development and maintaining physical and mental health. Sleep disturbances have long been known to be associated with mental illness, metabolic, neurological or other systems diseases. Knowing what factors are associated with sleep quality and sleep-wake homeostasis is important for the study of sleep disorders and may potentially inform new treatment strategies to preserve patients' normal sleep-wake cycle. The present study aims to identify what actigraphic measures, self-reported sleep variables, and other chronic diseases, medications are related to the percentage of slow-wave sleep and delta power spectra in older men.
Method: Categorical variables are summarized using frequencies and percentages. For continuous variables, means and standard deviations are computed, and distributions are displayed in histograms. Possible correlations among variables are examined by a matrix of scatterplots and Pearson correlation coefficients. The LASSO is used for feature selection in multiple linear regression models and multiple imputation used to overcome missing data.
Results: The past month sleep hours (β=0.0896, p<0.05), kidney diseases (β=0.161, p<0.05) and oral corticosteroids (β=0.148, p<0.05) are significantly positively associated with percentage of deep sleep, while sleep apnea severity (β=-0.0043, p<0.001), age ( = -0.0042, p<0.01), Benzodiazepine use ( -0.155, p<0.001), NSAIDS use (β=-0.0418, p<0.05), and race(β=-0.0476, p<0.01) are negatively associated when controlling other variables’ effect. Cognitive function (β=0.0015, p<0.001), and oral corticosteroids (β=0.0733, p<0.01) are positively related to delta power, while sleep apnea severity (β=-0.0011, p<0.001), age ( = -0.0013, p<0.05), mean sleep minutes (-0.0002, p<0.001) , BMI (-0.031, p<0.001), Diabetes (β=-0.0404, p<0.001), Benzodiazepine use ( -0.061, p<0.001), and the consumption of alcoholic beverages (β=-0.0125, p<0.05) are negatively related to delta power when controlling other covariates.
Conclusions: Our study suggested several factors are either positively or negatively associated with the percentage of deep sleep and delta power. Most of the factors affect the percentage of slow-wave sleep and delta power in the same direction.
Public Health Significance: These analyses may provide important messages for future study and potential medical interventions application.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Thesis AdvisorKrafty,
Committee MemberSmagula,
Committee MemberChang, Chung-Chou
Date: 26 September 2019
Date Type: Publication
Defense Date: 25 July 2019
Approval Date: 26 September 2019
Submission Date: 23 July 2019
Access Restriction: 1 year -- Restrict access to University of Pittsburgh for a period of 1 year.
Number of Pages: 54
Institution: University of Pittsburgh
Schools and Programs: School of Public Health > Biostatistics
Degree: MS - Master of Science
Thesis Type: Master's Thesis
Refereed: Yes
Uncontrolled Keywords: Slow-wave sleep, delta power spectra, sleep, statistical learning, Multiple regression, LASSO
Date Deposited: 26 Sep 2019 16:50
Last Modified: 01 Sep 2020 05:15


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item