Biostatistics Commons^™

Database-Driven Revelations In Epilepsy: Patient Reporting Accuracy And The Circadian Timing Of Seizures, Ithay Biton

Theses and Dissertations

For over thirty years, patients have been visiting the Arkansas Epilepsy Program for diagnosis and treatment for seizures and seizure-like episodes. As part of their clinical evaluation, patients often undergo ambulatory EEG (electroencephalogram) monitoring. This routine process produces valuable data for treating the patient. In this study, over 2000 ambulatory EEG reports from 1998 to 2016 were reviewed. A large database of seizures was created from the reports, with information on 407 patients, 1611 EEG-confirmed seizures, and 1726 patient-reported seizures (IRB Protocol #17-093). The database was used to address two important issues in epilepsy. The first topic of the study …

Type Ii Diabetes Treatment Comparison Via Compartment Modeling, Abigail M. Collins

Theses and Dissertations

Type II diabetes mellitus affects one in ten adults worldwide, yet the effects of treatment type and adherence level on developing complications and quality of life have not been well characterized at the population level, and mathematical modeling offers a structured way to examine these dynamics. This thesis adapts the Boutayeb et al. (2004) model to incorporate dynamic treatment types and levels of adherence, producing nine scenarios in which complication development rate and complication recovery rate differed, to compare peak complications and quality of life across treatment and adherence conditions. Using a system of ordinary differential equations and compartment modeling, …

Propensity Score Matching Accounting For Longitudinal Trends Before Baseline With Group-Based Trajectory Modeling, Dustin R. Bastaich

Theses and Dissertations

Propensity score matching is used in observational studies to balance baseline attributes between a treatment of interest and a control group. Propensity score matching typically relies on baseline variables, but longitudinal trends in patient characteristics can also influence treatment decisions and subsequent health outcomes. This dissertation extends standard approaches by explicitly incorporating longitudinal trajectories of key variables into the propensity score estimation process.

Trends in a longitudinal variable prior to baseline were characterized using group-based trajectory modeling. A two-step modeling approach was implemented where trajectory groups of a key variable were first estimated and then included as covariates in the …

Sequential Data Modeling Of Influenza A Via Traditional, Dwt-Gpr Hybrid, And Deep Learning Architectures, Edmund Fosu Agyemang

Theses and Dissertations

Influenza A is responsible for 290,000 to 650,000 respiratory deaths a year, though this estimate is an improvement from years past due to improved sanitation, healthcare practices, and vaccination programs. In this study, we perform a comparative analysis of traditional, deep-learning and discrete wavelet (DWT)-Gaussian Process (GP) hybrid models to predict Influenza A outbreaks. Using historical data from January 2009 to December 2023, we compared the performance of traditional ARIMA and ETS models, four variants of DWT-GPR models and six distinct deep learning architectures: Simple RNN, LSTM, GRU, BiLSTM, BiGRU and Transformer. The results reveal a clear superiority of all …

Approaches To Enhancing Multiple Hypothesis Testing Methods With Side-Information, Siyu Zheng

Theses and Dissertations

Lesion-symptom mapping (LSM) studies offer insight into the brain areas involved in various aspects of cognition. This is commonly done via behavioral testing in patients with a naturally occurring brain injury or lesions (e.g., strokes or brain tumors). This results in high-dimensional observational data where lesion status (present/absent) is non-uniformly distributed, with some voxels having lesions in very few (or no) subjects. In this situation, mass univariate hypothesis tests have severe power heterogeneity where many tests are known a priori to have little to no power. Additionally, high-dimensional observational data can be grouped according to brain anatomical structure.

In this …

A Spatial Scan Statistic For Group Testing Data, Vincent Onyame

Theses and Dissertations

Group testing involves pooling specimens from multiple individuals and offers an efficient means to surveil low-prevalence pathogens, but poses challenges for spatial cluster detection when only pooled results are observed. In this thesis, we develop a spatial scan statistic tailored to group-testing data with variable pool sizes. The statistic compares a null hypothesis of a homogeneous infection rate across all clusters to an alternative hypothesis that infection probabilities differ inside and outside a candidate cluster, with both models fitted by maximum likelihood estimation. We approximate the null distribution of the maximum likelihood ratio test via Monte Carlo simulation.

Through a …

Incorporating Propensity Score Weighting And Nonresposne Adjustments Into Complex Survey Data With Survival Outcomes, Xinrui Shi

Theses and Dissertations

Propensity score weighting (PSW) plays a key role in minimizing confounding in observational research, especially when estimating treatment effects for time-to-event outcomes. However, its integration into survey data with complex design – particularly data with multiple stage sampling and censoring – remains underexplored. One significant challenge in such settings is the presence of nonresponse, which can introduce additional bias and complicate the use of standard weight adjustments. Moreover, there has been limited study on how PS weights can be effectively combined with nonresponse weighting adjustments in complex survey data that include survival outcomes. This dissertation aims to extend current methodologies …

Joint Modelling Of Longitudinal Egfr Trajectory And Time To Acute Kidney Injury In Lung Transplant Patients, Samiha Zakir

Theses and Dissertations

Progressive declines in estimated glomerular filtration rate (eGFR) often precede acute kidney injury (AKI), yet the relationship between eGFR trends and AKI risk remains unclear. This study investigates longitudinal eGFR changes and their association with AKI in 459 lung transplant patients followed for up to 7 years (n = 6419). We applied a piecewise linear mixed-effects model to evaluate eGFR trajectories and a Cox proportional hazards model to assess time to AKI. A joint model was used to explore the interplay between longitudinal and survival processes. Key covariates included gender, age at transplantation, antibody-mediated rejection (AMR), and pre-transplant eGFR. Males …

Beyond Homogeneity: Exploring Causal Heterogeneity In Psychopathology, Philip B. Vinh

Theses and Dissertations

Traditional models in psychiatric research often impose assumptions of causal homogeneity, treating population-level associations as reflective of uniform underlying mechanisms. This dissertation challenges that assumption by introducing statistical and machine learning frameworks designed to detect and model causal heterogeneity in the development of psychopathology. Central to this approach is the advancement of finite mixture structural equation modeling (FM-SEM) to identify latent subgroups characterized by distinct, and sometimes opposing, causal pathways.

The dissertation comprises three integrated empirical studies. The first introduces mixDoC, a finite mixture extension of the classical Direction of Causation (DoC) model applied to twin data, enabling the detection …

An Adaptive Method For Covariate Balancing In Block Randomized Clinical Trials, Ren Rasnick

Theses and Dissertations

Clinical trials are randomized in part to limit allocation bias, but also to ensure comparability between treatment arms for a baseline variable of concern. Comparable with regard to a baseline variable of concern is necessary for the validity of statistical methods. However, comparability is not guaranteed for trials of any size and is even more likely in trials with < 200 total participants. We propose a new method for adapting the allocation of participants in a sequentially allocated two-armed study with a small sample size to better ensure comparability.

The proposed method calculates the expected final imbalance (lack of comparability) based on the current participant values. Unlike several other methods, our method ensures the final desired sample size for each treatment arm, utilizes the expected final imbalance, increases comparability between …

Predictive Modeling For Healthcare Data Using Nonlinear Bayesian Methods, Prince Kofi Asare

Theses and Dissertations

Unplanned hospital readmissions represent a significant challenge for healthcare systems, contributing to substantial financial burdens and highlighting gaps in patient care coordination. In the U.S., approximately 20% of Medicare beneficiaries are readmitted within 30 days, costing billions annually. Social determinants of health, such as income, housing stability, and social support, account for up to 80% of health outcomes, yet their integration into predictive models remains underexplored. This study introduces a novel Bayesian framework for predicting 30-day readmission risk, combining Gaussian Process models with spike-and-slab priors and Bayesian Lasso regression with Laplace priors. Utilizing Markov Chain Monte Carlo methods, the approach …

Method For Measuring The Rate Of Improvement In Survival Times Of Cancer Patients, Thobani Chaduka

Theses and Dissertations

In clinical settings, technological advancements have facilitated health care improvements in data analytics, artificial intelligence, telemedicine, health information systems, etc. This has furthered our understanding of cancer biology and treatment mechanisms. In this study, we aim to understand whether Moore’s law-like models may derive from historical cancer survival data, and how they can predict survival statistics for newly diagnosed cancer patients. Historically these predictions have previously been done with the diagnosis year as the independent variable and the survival as a dependent variable. In this study we use death year data as an independent variable and from that, we derive …

Simulation Study On Count Data Based On Double Poisson Distribution, Chao Ma

Theses and Dissertations

This thesis delves into the double Poisson distribution. Regression based on the double Poisson distribution, as proposed by Efron in 1986, offers an alternative approach that allows for more accurate regression models when dealing with discrete data that exhibit either over- or under-dispersion compared to the Poisson distribution. In this thesis, two methods of calculating the exact double Poisson density are compared: one utilizes the exact probability with the normalizing constant c(mu, theta) by definition or the “finite sum” method, while the other employs an approximation of the normalizing constant c(mu, theta). Furthermore, a simulation was …

Evaluating The Influence Of Perfluorooctane Sulfonic Acid Exposure On Blood Glucose Levels: A Comprehensive Multiple Regression Analysis Considering Confounding Factors, Henry Mensah

Theses and Dissertations

Perfluorooctane sulfonate (PFOS) are widely used for industrial and commercial purposes and have received increasing attention due to their adverse effects on health. This thesis investigates the relationship between PFOS exposure and blood glucose levels, considering potential confounding factors. Regression analysis was conducted on a dataset comprising demographic, lifestyle, and biomarker data from a diverse population sample. Sex exhibited a significant role, with females demonstrating an 11.77% increase in blood glucose levels in response to PFOS exposure compared to males, supported by a p-value of 1.04 × 109. Body mass index (BMI) also played a pivotal role, revealing a 2.17% …

The Genetic Architecture Of Cervical Change During Pregnancy: From Modeling To Mechanism — Does The Cervix Mediate Maternal Risk For Spontaneous Preterm Birth?, Hope M. Wolf

Theses and Dissertations

This project leverages clinical data and biospecimens from a prospective longitudinal cohort of pregnant women to study the genetic and phenotypic relationships between cervical shortening and the duration of pregnancy. Sonographic cervical length (CL) was measured throughout pregnancy in a cohort of 5,160 Black/African American women in Detroit, Michigan. Maternal DNA samples were sequenced with a next-generation low-pass whole genome platform. The heritability of cervical change during pregnancy and its genetic correlation with gestational age at delivery (GAD) were estimated using Genome-Wide Complex Trait Analysis. These estimates suggest that cervical change is heritable (h²CL = 51%) and highly polygenic trait. …

Approaches To Detecting And Modeling Over-And Underdispersion In Alternative Count Data Distributions And An Application Of Logistic Regression And Random Forest Modeling To Improve Screening Tools For Tic Disorders In Children, Rebecca C. Wardrop

Theses and Dissertations

This dissertation focuses on theory and application of discrete data methods, particularly approaches to over- and underdispersion relative to the Poisson distribution and an application of random forest and logistic regression modeling. The first chapter derives a score test for over- and underdispersion in the heaped generalized Poisson distribution. Equi-, over-, and underdispersed heaped generalized Poisson and heaped negative binomial data are simulated to evaluate the performance of the score test by comparing the power it achieves to that of Wald and likelihood ratio tests. We find that the score test we derive performs comparably to both the Wald and …

Statistical Methods For Single Cell Sequencing Data Analysis, Fei Qin

Theses and Dissertations

The recent emergence of single cell sequencing (SCS) technology has provided us with single-cell DNA or RNA sequencing (scDNA/RNA-seq) information to investigate cellular evolutionary relationships. Despite many analysis methods have been developed to infer intra-tumor genetic heterogeneity, cluster cellular subclones, detect genetic mutations, and investigate spatially variable (SV) genes, exploring SCS data remains statistically challenging due to its noisy nature.

To identify subclones with scDNA-seq data, many existing studies use an independent statistical model to detect copy number profile in the first step, followed by classical clustering methods for subclone identification in downstream analyses. However, spurious results might be generated …

A Bayesian Spatial Scan Statistic For Normal Data, Laasya Velamakanni

Theses and Dissertations

Scan statistics are useful methods for detecting spatial clustering. While they were initially developed to detect regions with an excess of binomial or Poisson events, spatial scan statistics have been extended to detect hotspots in other types of data including continuous data. They have many applications in different fields such as epidemiology (e.g. detecting disease outbreaks), sociology (e.g. detecting crime hotspots), and environmental health (e.g. detecting high-pollution areas). Spatial scan statistics identify a ‘most likely cluster’ and then use a likelihood ratio test to determine if this cluster is statistically significant. Spatial scan statistics have been extended to the Bayesian …

Detecting Gene-Gene Or Gene Environment Interactions In Association With Complex Disease Outcomes, Taiwo Adetunji Famuyiwa

Theses and Dissertations

Identifying gene–gene and gene–environment interaction is complicated and gainsaying most especially because of high multifactor dimensions involve in the analysis of such combination discrediting the functionality of parametric statistical method like Logistic Regression. [22] developed a multi-factor-dimensionality reduction (MDR) method for detecting and characterizing high-order gene-gene and gene-environment interactions in case-control and discordant-sib-pair studies with relatively small samples, which is inspired by the combinatorial partitioning method of [18]. [16] proposed Generalized MDR (GMDR) framework based on the score of a generalized linear model which allows adjustment of covariates, provides a unified framework for handling both dichotomous and quantitative phenotypes. However, …

Identifying Limitations In Using Diagnostic Testing For Absorption Of Passive Maternal Immunity In Neonatal Beef Calves To Predict Pre-Weaning Disease, Alexis Charlotte Thompson

Theses and Dissertations

Calves are born agammaglobulinemic and rely on colostrum consumption for the transfer of maternal passive immunity. Calves that fail to absorb adequate amounts of maternal antibodies from colostrum are commonly referred to as having failed transfer of passive immunity (FTPI). The overall aim of this dissertation was to explore the usefulness of FTPI testing in neonatal beef calves to predict their risk for subsequent illness or death. The objectives were to evaluate the impact of FTPI on pre-weaning disease in beef and dairy calves, quantify and compare the variance in IgG concentrations measured by radial immunodiffusion and serum total protein …

Survival Models With Background Mortality, Shujie Chen

Theses and Dissertations

In this dissertation, we focus on studying three mixture cure models with background mortality. With the development of treatment, patients may be cured and suffer from other cause of death. The cure model with background mortality can measure the population cure which refers to the patients with comparable mortality with their counterpart in general population. Three types of survival models are investigated, including generalized odds rate (GOR) model, cure model with background mortality for right censoring and interval censoring, and extended illness death model via incorporating “cure” fraction. All methods are validated via comprehensive simulation studies and real data application. …

Detecting Spatially Varying Coefficient Effects With Conditional Autoregressive Models: A Simulation Study Using Social Determinants Of Health Screening Data, Reid J. Demass

Theses and Dissertations

Generalized linear models which include spatially varying coefficient terms allow researchers to determine if the association between predictor and outcome variables vary across geographic space. Such models are particularly applicable to research with public health data where interventions and limited health care resources must be allocated carefully. The integrated nested Laplace approximation (INLA) methodology available in the R INLA package is a popular tool to estimate spatially varying coefficients. To assess the performance of the estimation procedure, patient emergency department (ED) visits were simulated from data sourced from a pilot study at Prisma Health. The INLA technique was used to …

Sparse Partitioned Empirical Bayes Ecm Algorithms For High-Dimensional Linear Mixed Effects And Heteroscedastic Regression, Anja Zgodic

Theses and Dissertations

Variable selection methods in both the frequentist and Bayesian frameworks are powerful techniques that provide prediction and inference in high-dimensional linear regression models. These methods often assume independence between observations and normally distributed errors with the same variance. In practice, these two assumptions are often violated. To mitigate this, we develop efficient and powerful Bayesian approaches for linear mixed modeling and heteroscedastic linear regression. These method offers increased flexibility through the development of empirical Bayes estimators for hyperparameters, with computationally efficient estimation through the Expectation Conditional-Minimization (ECM) algorithm. The novelty of these approaches lies in the partitioning and parameter expansion, …

Variability In Causal Effects On A Binary Outcome And Noncompliance In A Multisite Randomized Trial, Xinxin Sun

Theses and Dissertations

Noncompliance to treatment assignment is widespread in randomized trials and presents challenges in causal inference. In the presence of noncompliance, the most commonly estimated effect of treatment assignment, also known as intent-to-treat (ITT) effect, is biased. Of interest in this setting is the complier average causal effect (CACE), the ITT effect among compliers. Further complication arises when the outcome variable is partially observed.

My research focuses on estimating the distribution of a site-specific CACE in a multisite randomized controlled trial (MRCT) by maximum likelihood (ML). Assuming compliance missing at random (MAR). We express the likelihood as an integral with respect …

Reassessing Replication: Addressing The Replication Crisis From A Statistical Perspective, Alicia Richards Phd

Theses and Dissertations

In 2015, Open Science Framework directly replicated 100 psychology studies and found astonishingly low replication rates. Since, researchers have suggested factors that may have influenced the low rates, including the metrics used to assess replications. The definitions used to decide whether a replication study was successful all suffer from flaws. Therefore, we propose a new metric for assessing replication that can estimate the likelihood a study successfully replicated rather than forcing a binary choice and accounts for study design limitations.

Using equivalence study techniques, we first propose a new metric to assess replication, defining a successful replication as one where …

Model-Based Imputation Of Below Detection Limit Missing Data And Group Selection In Bayesian Group Index Regression, Matthew Carli

Theses and Dissertations

Investigations into the association between chemical exposure and health outcomes are increasingly focused on the role of chemical mixtures, as opposed to individual chemicals. The analysis of chemical mixture data required the development of novel statistical methods, one of these being Bayesian group index regression. A statistical challenge common to all chemical mixture analyses is the ubiquitous presence of below detection limit (BDL) data. We propose an extension of Bayesian group index regression that treats both regression effects and missing BDL observations as parameters in a model estimated through a Markov Chain Monte Carlo algorithm that we refer to as …

Integrative Post-Gwas Analyses Of Psychiatric Disorders: Identifying Putative Risk Genes And Gene Sets Using Transcriptome, Proteome And Methylome Information, Huseyin Gedik

Theses and Dissertations

Genome-wide association studies (GWAS) of psychiatric disorders (PD) yield numerous loci with significant signals, but often they do not implicate specific protein coding genes. Because GWAS risk loci are enriched in expression/protein/methylation quantitative loci (e/p/mQTL, hereafter xQTL), transcriptome/proteome/methylome-wide association studies (T/P/MWAS, hereafter XWAS), which integrate information from GWAS and x-level (mRNA, protein or DNA methylation levels) coming from largest xQTL studies, can link GWAS signals to effects on specific genes. For gene level analyses, researchers use mendelian randomization (MR) methods to fine-map the association between x-levels and trait. However, none of the previous studies ever jointly analyzed XWAS of multiple …

Dynamics Of Redox-Driven Molecular Processes In Local And Systemic Plant Immunity, Philip Berg

Theses and Dissertations

The work here presents two main parts. In the first part, chapters 1 – 3 focus on dynamical systems modeling in plant immunity, whereas chapters 4 – 6 describe contributions to computational modeling and analysis of proteomics and genomics data. Chapter 1 investigates dynamical and biochemical patterns of reversibly oxidized cysteines (RevOxCys) during effector-triggered immunity (ETI) in Arabidopsis, examines the regulatory patterns associated with Arabidopsis thimet oligopeptidase 1 and 2’s (TOP1 and TOP2), roles in the RevOxCys events during ETI, and analyzes the redox phenotype of the top1top2 mutant. The second chapter investigates the peptidome dynamics during ETI …

Towards Structured Planning And Learning At The State Fisheries Agency Scale, Caleb A. Aldridge

Theses and Dissertations

Inland recreational fisheries has grown philosophically and scientifically to consider economic and sociopolitical aspects (non-biological) in addition to the biological. However, integrating biological and non-biological aspects of inland fisheries has been challenging. Thus, an opportunity exists to develop approaches and tools which operationalize planning and decision-making processes which include biological and non-biological aspects of a fishery. This dissertation expands the idea that a core set of goals and objectives is shared among and within inland fisheries agencies; that many routine operations of inland fisheries managers can be regimented or standardized; and the novel concept that current information and operations can …

Topics In Multilevel Mediation Analysis, Chung Li Wu

Theses and Dissertations

A proper study design assures adequate power to detect statistically significant differences. Existing power calculations for multilevel mediation analysis make a strong distributional assumption of normality. However, binary outcomes are commonly seen in real-world study. Motivated by this fact, we conduct a simulation-based power study for a multilevel mediation analysis with binary outcomes. The numbers of participants needed to achieve 80% power are summarized in tables for future reference.

Mixed-effect regression is commonly used in multilevel analysis for panel data. Yet, the estimated coefficients from the random-intercept model could represent either purely between-cluster, purely within-cluster, or weighted-average effects. Therefore, we …