Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


33097.pdf (19.9 MB)

ETD Abstract Container

Abstract Header

Bayesian Shape Invariant growth curve model for longitudinal data

Bhuiyan, Mohammad AN
http://orcid.org/0000-0002-6011-2624
http://rave.ohiolink.edu/etdc/view?acc_num=ucin1561393650064101

Abstract Details

2019, PhD, University of Cincinnati, Medicine: Biostatistics (Environmental Health).
The analysis of growth curves has played a vital role in estimating the growth trajectory of populations as well as identifying critical factors corresponding to various shapes of those trajectories. In recent years, shape invariant modeling has become an active area of research for non-parametric growth curve modeling, where a single function is transformed by scaling and shifting it to fit each subject usually through affine transformations. Lawton, first proposed SIM called it self-modeling regression; in their approach, the function for the underlying shape is illustrated for various parametric functions. Later, Beath developed a model to explain longitudinal growth patterns and extended the SIM to include time-dependent covariates. As a type of SIM, the regression spline expressed as a basis function consisting of a different set of knots; the resulting structure fitted as a nonlinear mixed effects model and parameters are typically estimated using maximum likelihood. This allows estimating the parameters for the between-subjects variation. The research in longitudinal growth curve modeling utilizing the Bayesian inferential procedure is limited, and the wider application is hindered by the computational complexities involved in such models. Cole proposed Super Imposition by Translation and Rotation model and expressed individual growth curves through three subject-specific parameters; named as size, tempo, and velocity. It is an important inferential problem to test no association between two binary variables based on data. A Test-based on sample odds ratio is commonly used. We bring in a competing test based on the Pearson correlation coefficient. An Odds ratio does not extend to higher order contingency tables, whereas the Pearson correlation does. It is a useful exercise to understand how the Pearson correlation stacks against the odds ratio in 2x2 tables. Another measure of association is the canonical correlation. In my second chapter we used power comparisons in 2x2 Contingency Tables: Odds Ratio versus Pearson Correlation versus Canonical Correlation to understand how Pearson correlation stacks against the odds ratio in 2x2 tables in the test of association. Air pollution is a growing global challenge and may have a moderate to the severe negative impact on human health. Vehicles, households, and industries emit a complex mixture of air pollutants, within which ambient particulate matter smaller than 2.5 micro m PM2.5 are thought to have the greatest effect on human health. Prior epidemiologic evidence suggests short-term PM2.5 exposure is associated with the development and exacerbation of several health problems. Children are more susceptible to PM2.5 related health effects due to their immature immune system and ongoing development and growth. The relationship of PM2.5 with asthma emergency department visit between 2011 and 2015 was identified within the Cincinnati Children’s Hospital Medical Center electronic medical record based on International Classification of Disease (ICD-9) and we used a data-driven clustering algorithm to find any clustering patterns existed by day within the study duration. In finding the impact of PM2.5 on stoke. we used a case-crossover design, to examine the association of exposure to PM2.5 and onset of incident stroke for the calendar year 2010.
Marepalli Rao, Ph.D. (Committee Chair)
Monir Hossain, Ph.D. (Committee Member)
Jane Khoury, Ph.D. (Committee Member)
Heidi Sucharew, Ph.D. (Committee Member)
Rhonda Szczesniak, Ph.D. (Committee Member)
Jessica Woo, PhD (Committee Member)
132 p.
Biostatistics
shape invariant; longitudinal; Growth curve; case-crossover

Recommended Citations

Citations

  • Bhuiyan, M. A. (2019). Bayesian Shape Invariant growth curve model for longitudinal data [Doctoral dissertation, University of Cincinnati]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1561393650064101

    APA Style (7th edition)

  • Bhuiyan, Mohammad. Bayesian Shape Invariant growth curve model for longitudinal data. 2019. University of Cincinnati, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=ucin1561393650064101.

    MLA Style (8th edition)

  • Bhuiyan, Mohammad. "Bayesian Shape Invariant growth curve model for longitudinal data." Doctoral dissertation, University of Cincinnati, 2019. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1561393650064101

    Chicago Manual of Style (17th edition)

ucin1561393650064101
94
© 2019, all rights reserved.
This open access ETD is published by University of Cincinnati and OhioLINK.
Release 3.2.12