Doctoral Dissertations

Luyang Wang

Abstract

”Sleep is one of the most important behaviors in animals, yet many aspects of sleep are not well understood. In humans, sleep has been shown to impact different aspects of health and cognitive performance. The fruit fly, Drosophila melanogaster, can be used as a model organism to investigate sleep patterns over an entire lifespan. The sleep-wake status can be recorded every minute over the life of individual flies, resulting in a wealth of data for identifying meaningful aspects of sleep. In this work, a framework based on statistical modeling methods is developed for sleep data collected on fruit flies to identify a set of novel sleep characteristics that can predict fly lifespan. These models are used to predict an independent set of flies into long- and short-lived groups at around the average fly midlife (30 days). Aging markers can then be assessed for differences in the predicted groups to determine if the model based on sleep features aligns with expected molecular differences.

The statistical modeling framework consists of first defining a set of novel sleep variables based on 30 days of the minute-level sleep-wake data. These variables include the proportion of time asleep and transition probabilities of staying asleep or awake from one minute to the next. These probabilities are calculated on a daily basis that change across the life of a fly. To incorporate different aspects of stability of the variables over time, the average and standard deviation of the first differences in the daily values are calculated. Differences in the trajectory shape of these variables over time are captured using scores calculated from functional principal component analysis (FPCA). Once the sleep features are derived, the next step in the framework consists of using two different types of statistical models (multiple linear regression and Cox proportional hazards regression) paired with two different model selection methods (stepwise and lasso) to select a model that uses sleep features to predict lifespan. The framework also features a resampling and cross validation approach that aims to yield a robust model for evaluation in an independent set of flies”--Abstract, page iii.

Advisor(s)

Olbricht, Gayla R.

Committee Member(s)

Samaranayake, V. A.
Paige, Robert L.
Wen, Xuerong Meggie
Thimgan, Matthew S.

Department(s)

Mathematics and Statistics

Degree Name

Ph. D. in Mathematics

Comments

Doctor of Philosophy in Mathematics with Statistics Emphasis

Publisher

Missouri University of Science and Technology

Publication Date

Fall 2019

Pagination

x, 82 pages

Note about bibliography

Includes bibliographic references (pages 79-81).

Rights

© 2019 Luyang Wang, All rights reserved.

Document Type

Dissertation - Open Access

File Type

text

Language

English

Thesis Number

T 12061

Electronic OCLC #

1313117363

Download

Share

 
COinS