In the information age, bridging the gap between data and clinical application is crucial. The Center for Clinical and Translational Science (CCTS) is making significant investments in this area. Over the years, the CCTS has supported the development of statistical methods for parameter selection, non-constant variance, drug screening, DNA sequence analysis, and adaptive clinical trial design. While these projects addressed specific disease challenges, the methods developed have broad applications, particularly for health conditions disproportionately affecting populations in the Deep South.The CCTS congratulates the following investigators for their successful submissions to our Center’s 2024 Statistical and Analytic Methods Development program. Read more about their projects below and if you are interested in applying for the current funding cycle, learn more and apply by Feb. 25, 2025.
Principle Investigator: Loren Gragert
Tulane UniversitySchool of Medicine, Department of Medicine
Division of Biomedical Informatics and Genomics.
Title of Project 1: Profiling performance of high-resolution Human Leukocyte Antigen (HLA) imputation with complementary metrics and visualizations using an open-source validation framework
Title of Project 2: Modeling Impact of Probabilistic Offer Filters based on High Resolution HLA Imputation to Reduce Histocompatibility-Related Organ Offer Refusals
Allocation of healthcare resources is a translational barrier affecting many health conditions and outcomes. One of the most recognized limited resources is donor-derived tissues and organs for transplantation. Development of allocation policy involves balancing factors such as medical urgency, post-transplant survival, candidate biology, patient access, and placement efficiency. Dr. Loren Gragert, PhD, Assistant Professor in the Division of Biomedical Informatics and Genomics at Tulane University focuses on improving how the allocation system captures data on immune compatibility and uses that data to determine to whom offers of specific organs should be made. Dr. Gragert secured two CCTS Methods Development awards in 2024, both focusing on compatibility assessments. His first CCTS award supported the development of a validation framework to show that molecular mismatch metrics that would improve post-transplant survival could be determined with high confidence if molecular typing data were captured in the system. After reporting significant progress, Dr. Gragert secured a second round of CCTS support that would study the impact that use of molecular typing data in allocation would have on reducing organ offer refusals related to immune compatibility. This clinical/translational work will illustrate that the development of clinical informatics tools and data standards can increase the precision of immune compatibility assessments and improve the utility and efficiency of the organ allocation system.
Principle Investigator: Evrim Oral
Associate Professor, School of Public Health, Department of Biostatistics and Data ScienceLouisiana State University Health Sciences Center
Title of Project: Poisson Regression with Stochastic Covariates for Environmental Data Modeling
Drawing evidence-based conclusions requires a systematic approach, involving careful study design and thorough data analysis. While this may seem straightforward, the process is often more complex. Statisticians play a crucial role, continuously refining and developing methods to ensure accurate statistical inferences. Dr. Evrim Oral, an Associate Professor in the Department of Biostatistics and Data Science at the Louisiana State University Health Sciences Center (LSUHSC), identified a statistical gap during her research. Traditional statistical models assume that the variables we study (known as "covariates") are fixed or predetermined. However, in observational studies, data comes from a random sample of measurements and individuals, over which we have no control. Furthermore, traditional models assume these variables follow a normal distribution, meaning there are equal numbers of low and high values, with the majority of measurements clustering around the average. Many studies, however, involve non-normal data. For example, consider assessing the relationship between air quality measurements and emergency room (ER) visits. Both air quality and ER visits exhibit considerable variability, and traditional statistical models are unlikely to accurately estimate the true relationship. To address this challenge, Dr. Oral engaged the CCTS to support the development of a novel statistical method designed to analyze count-based data (e.g., the count of days per month with high particulate matter, the count of emergency room visits per day) while accounting for covariates that are non-fixed and highly variable. Although Dr. Oral plans to apply this model to environmental data, it will be highly valuable across many fields of research that deal with count-based datasets containing covariates with significant outliers.
Principle Investigator: Yipu Zhang
Assistant Professor, School of Medicine, Department of Biomedical Informatics and GenomicsUniversity of Alabama at Birmingham
Title of Project: A Privacy-Preserving Multi-Site MRI Analysis for Alzheimer's Disease and Cognitive Function
In clinical research, gaps in data management can significantly impact study operations, from regulatory compliance to data sharing. This issue is especially pronounced in clinical imaging, where variations in imaging acquisition methods, large file sizes, and privacy concerns hinder data analysis and multi-site studies. Dr. Yipu Zhang, Assistant Professor at the School of Medicine, Department of Biomedical Informatics and Genomics, indicates that multi-site clinical imaging data is prone to systematic biases, sub-par data labeling, and imbalanced sample sizes. To address these challenges, Dr. Zhang will leverage clinical imaging data from multiple national-level clinical studies to develop and test a federated framework. This system allows data processing across multiple devices or locations, enhancing model training while ensuring privacy. To encourage broader adoption of the framework, Dr. Zhang also plans to release an open-source toolbox. While the current work focuses on scoring MRI-based brain scans from patients with Alzheimer’s Disease, the approach is applicable and scalable for multi-center and multi-modal MRI data supporting psychiatric and neurological imaging studies. Dr. Zhang’s work highlights how evolving data management approaches enhance the rigor, reproducibility, and transparency in the conduct of clinical research.
The CCTS Statistical and Analytic Methods Development opportunity is funded via the NIH (Grant Number: UM1TR004771).