Research Projects
Statistical Methods for Kidney Markers as Shared Determinants of Dementia and Physical Disability in Older Adults (PI: M Shardell)
Beyond known kidney effects on bone, declining kidney health may lead to dementia and physical disability. Therefore, this study aims to determine whether kidney markers representing multiple aging-related biological mechanisms affect relations and dynamics between cognitive and physical endpoints. We accomplish this goal using methods that address time-varying confounding and truncation due to death.
Methods to Test Biomarkers of Aging as Shared Determinants of Alzheimer’s Disease and Related Dementias and Physical Disability (PI: M Shardell)
The goal of this project is to identify early biomarkers of biological aging mechanisms that are related to dual cognitive-physical decline and joint ADRD-disability onset in initially healthy older adults. This is accomplished by extending statistical methods for multivariate longitudinal and time-to-event outcomes and applying them to harmonized data from 8 cohort studies of >11,000 community-dwelling adults aged at least 65 years.
Methods to Test the Role of Age-related Lifestyle and Vaginal Microenvironment Changes and the Prevention, Treatment, and Progression of Genitourinary Syndrome of Menopause (PI: M Shardell/R Brotman)
This project aims to identify core age-related vaginal microenvironment biomarkers (microbial, metabolite, and immune profiles) and determine how they affect the genitourinary syndrome of menopause. We accomplish this goal by carrying out a large longitudinal study of women across reproductive stages. We simultaneously address both conventional epidemiologic challenges (missing data, confounding) and methodological challenges arising from high-dimensional multi-omic data.
Developing a Novel Analytical Toolbox to Tackle Multifaceted Statistical Challenges in Analyzing Post-Fracture Recovery Trajectories in Older Adults with ADRD (PI: C Chen)
This project involves developing, validating, and applying novel analytical methods in data science, including proposing machine-learning-assisted high-dimensional regression, computationally efficient individualized dynamic prediction, and multi-algorithm-based robust causal inference methods. Methods are applied to a dataset comprising >20,000 Medicare beneficiaries treated at over 1000 hospitals to understand multilevel variability in post-fracture recovery outcomes for older adults living with ADRD.
A BRAIN Initiative Resource: The Neuroscience Multi-omic Data Archive (PI: O White/A Mahurkar)
The Brain Research through Advancing Innovative Neurotechnologies (BRAIN) Initiative promotes the development and application of technologies to describe the temporal and spatial dynamics of cell types and neural circuits in the brain. To promote smooth interactions across a large research consortium, we have developed the Neuroscience Multi-Omic Archive (NeMO Archive), a data repository focused on the storage and dissemination of omic data from the BRAIN Initiative and related brain research projects.
NeMO Archive: SCORCH Support, Coordination and Outreach (PI: O White/S Ament/A Mahurkar)
This project leverages the Institute for Genome Sciences Neuroscience Multi-Omic Data Archive to establish a data coordination center to analyze single cell and other molecular data sets generated by Single Cell Opioid Responses in the Context of HIV (SCORCH) and other NIDA-funded HIV and substance use disorder projects. Notably, this project represents several key existing resources for data management, integration, and web presentation tools developed by the University of Maryland group.
A consortium facilitation, coordination, and data management center for the HVP (PI: O White/C Huttenhower)
The Human Virome Project (HVP) is an ambitious initiative aimed at characterizing the lesser- explored communities of bacteriophage and eukaryotic viruses associated with the human body, collectively known as the virome. This project establishes a collaborative and state-of-the-art Consortium Organization and Data Collaboration Center (CODCC) for the HVP. Leveraging the team's expertise and resources from projects such as the Human Microbiome Project and the Integrative Human Microbiome Project, the CODCC is designed to be experienced, scientifically knowledgeable, and administratively and computationally efficient.
Alternative Stimulation Mode and Location for Auditory Hallucination Neuromodulation Treatment (PI: L Hong/S Chen/X Du)
This project is a two-phase randomized study to evaluate repetitive Transcranial Magnetic Stimulation (rTMS) treatment for schizophrenia spectrum disorder. We use a novel approach to design rTMS treatment for auditory hallucination that is based on a strong neural circuitry mechanism on auditory hallucination formation. The ultimate goal is to determine whether the new stimulation site and mode strategy indeed significantly engages the proposed circuitry of action.
Analytic Support to the UMB Scientific Research Community
The Division of Biostatistics and Bioinformatics serves the University of Maryland, Baltimore, investigators and trainees by providing biostatistical collaboration and consultation services, and access to state-of-the-art statistical science, expert statistical analysis, and sound statistical reporting in biomedical research. The Division plays a pivotal collaborative role in all phases of biomedical research from initial study design to final project report. For example, biostatisticians provide consultation and collaboration on the research design and statistical analysis plan, including assessing study feasibility, estimating sample sizes to meet study objectives, and preparing grants and manuscripts. Examples of this service include:
Study Design and Power Analysis. Feasibility study and sample size based on state-of-the-art methodology such as multi-stage and adaptive designs. In addition, we design and perform randomization, sensitivity and simulation studies to evaluate study design properties, and sample sizes needed to achieve study objectives, endpoints and control selection.
Statistical Analysis. The statistical analysis plan is provided in the protocol (before data collection takes place) and followed. Assumptions underlying the methods are assessed and the most appropriate statistical method is used for analysis. Data analysis is performed in major statistical software packages such as SAS, Stata, R, Matlab, or Python.
Interim Analyses and Data and Safety Monitoring for Clinical Trials. Analyses and adaptive designs (including Bayesian designs) consistent with statistical principles for multiple testing and interim analyses are performed.
Bioinformatics and multi-omic data analysis. Another focus of the Division is bioinformatics and the analysis of high-dimensional multi-omic data including systems biology, gene regulatory network inference, module biomarker discovery, network biology, data integration, systems pharmacology, personalized medicine, disease-drug-gene association discovery, pattern recognition, image analysis and data mining.
Reports and Publications. Data analysis is summarized in a statistical report in a form appropriate for manuscripts.
Members of the University of Maryland Marlene and Stewart Greenebaum Cancer Center can request biostatistics and bioinformatics services here. Other investigators can access the service here. Questions can be directed to Susan Holt at sholt@som.umaryland.edu.