A joint modeling approach for longitudinal microbiome data with time-to-event outcomes
| dc.contributor.advisor | Scott, David W. | en_US |
| dc.contributor.committeeMember | Shaw, Chad A. | en_US |
| dc.creator | Luna, Pamela Nicole | en_US |
| dc.date.accessioned | 2019-05-17T19:03:20Z | en_US |
| dc.date.available | 2020-05-01T05:01:09Z | en_US |
| dc.date.created | 2019-05 | en_US |
| dc.date.issued | 2019-04-15 | en_US |
| dc.date.submitted | May 2019 | en_US |
| dc.date.updated | 2019-05-17T19:03:20Z | en_US |
| dc.description.abstract | Humans are living, breathing ecosystems. We share our bodies with a vast collection of microorganisms that easily outnumber the human cells in our bodies. While these microbes are mostly benign or beneficial, non-pathogenic and pathogenic microorganisms alike can be harmful in certain abundances. Microbial imbalances within anatomic sites have been linked to human illness, but the underlying mechanics of how individuals reach this state of dysbiosis are not well understood. Determining how changes in microbial abundances affect the onset of disease could lead to novel treatments that help to stabilize the microbiome. However, currently no methods appropriately quantify the associations between longitudinal microbial abundance changes and time-to-event outcomes. This lack of methodology is partially due to the inability to include time-dependent biomarker data in an event model. Though the existing joint model for longitudinal and time-to-event data addresses this issue by incorporating a linear mixed effects model into the hazard function of the Cox proportional hazards model, it does not allow for proper modeling of non-Gaussian microbiome data. In this thesis we present a joint modeling approach which uses a negative binomial mixed effects model to determine the longitudinal values included in the hazard function. We discuss how our model parameterization generates interpretable results and accounts for the underlying structure of microbiome data while still respecting its inherent compositionality. We outline how to simulate longitudinal microbiome data and associated event times to create a joint model dataset and show that our model performs better than existing alternatives. Finally, we illustrate how this methodology could be used to improve clinical diagnostics and therapeutics. | en_US |
| dc.embargo.terms | 2020-05-01 | en_US |
| dc.format.mimetype | application/pdf | en_US |
| dc.identifier.citation | Luna, Pamela Nicole. "A joint modeling approach for longitudinal microbiome data with time-to-event outcomes." (2019) Diss., Rice University. <a href="https://hdl.handle.net/1911/105994">https://hdl.handle.net/1911/105994</a>. | en_US |
| dc.identifier.uri | https://hdl.handle.net/1911/105994 | en_US |
| dc.language.iso | eng | en_US |
| dc.rights | Copyright is held by the author, unless otherwise indicated. Permission to reuse, publish, or reproduce the work beyond the bounds of fair use or other exemptions to copyright law must be obtained from the copyright holder. | en_US |
| dc.subject | Microbiome | en_US |
| dc.subject | longitudinal | en_US |
| dc.subject | time-to-event | en_US |
| dc.subject | joint modeling | en_US |
| dc.title | A joint modeling approach for longitudinal microbiome data with time-to-event outcomes | en_US |
| dc.type | Thesis | en_US |
| dc.type.material | Text | en_US |
| thesis.degree.department | Statistics | en_US |
| thesis.degree.discipline | Engineering | en_US |
| thesis.degree.grantor | Rice University | en_US |
| thesis.degree.level | Doctoral | en_US |
| thesis.degree.name | Doctor of Philosophy | en_US |
Files
Original bundle
1 - 1 of 1