Dissertations & Theses (Open Access)

Date of Award

Summer 8-10-2021

Degree Name

Master of Science (MS)


Ruosha Li, Ph.D.

Second Advisor

Stacia Desantis, Ph.D.

Third Advisor

Licong Cui, Ph.D.


Objectives: Pregnant women in the third trimester are more likely to experience sleep disorders, resulting in adverse effects on fetal health. This study aimed to evaluate the sleep efficiency (SE) of nulliparous women in their late pregnancy.

Methods: Data was extracted from the National Sleep Research Resource focusing on nulliparous women (n = 1,803) in their third trimester of pregnancy. The two-sample Wilcoxon test for continuous variables and the Chi-square test for categorical variables were used based on a normal SE cut-off of 0.85. The least absolute shrinkage and selection operator (Lasso) was used to find the risk factors of the nulliparous women affecting the SE. Model results and predictive performance were evaluated and compared using Lasso and machine learning methods.

Results: A total of 13.5 % of pregnant women in the study had poor sleep efficiency. Among them, 43.9% were White, 21.3% were Black, 24.6% were Hispanic, and 3.7% were Asian. The Lasso model selected total sleep duration with >92% oxygen saturation, number of sleep-to-awake shifts per hour, and number of desaturations with >= 2% oxygen desaturation in total sleep period as the most contributing variables. Random forest feature importance selected the same result as the Lasso feature selection on the top three variables.

Conclusions: The study showed that the sleep efficiency of nulliparous women in the third trimester was significantly associated with oxygen saturation and desaturation status as well as change in sleeping patterns after sleep onset. Reduced time spent supine may be favourable to the breathing condition of pregnant women, thus improving their sleep efficiency. The Lasso model can be powerful in narrowing down useful information from a large number of variables.

Included in

Public Health Commons