Category: Research, Design, and Statistics

What if We Ignore the Random Effects When Analyzing RNA-seq Data in a Multifactor Experiment?

Identifying differentially expressed (DE) genes between different conditions is one of the main goals of RNA-seq data analysis. Although a large amount of RNA-seq data are produced for two-group comparison with small sample sizes at early stage, more and more RNA-seq data are being produced in the setting of complex experimental designs such as split-plot […]

Robust Estimation of Propensity Score in Observational Study

Observational Study aims to draw inference about the possible effect of a treatment on subjects from an empirical comparison of treated and controlled groups and is widely used in healthcare and medical research when clinical trials are not applicable in practice. The central issue in observational study is to identify potential confounding to ensure the […]

Time Series Analysis of 30-Day Readmission Rates: Health Care Innovation – Bridging the Divide

Application of times series methods to 30-day readmission rates of PCI patients in the BRIDGES database. Methods include interrupted time series, cross-correlation and Granger causality. Dr. Paul Kolm, DAC-CTR Associate Director, is Director of Biostatistics at Christiana Care Health System, and Research Professor of Medicine at Thomas Jefferson Medical College. He has been and is […]

Reducing Selection Bias Inverse Probability Weighting and Bin Bootstrapping

Reducing selection bias is a concern when analyzing observational datasets. IPW (inverse probability weighting) and PSBB (propensity score bin bootstrapping) are two methods used to address selection bias. This presentation will demonstrate how IPW and PSBB capture information from all patients with balance achieved for measured confounders via propensity score adjustment. Dr. Zugui Zhang is […]

Propensity Score Matching for Estimating Treatment Effects

Propensity Scores matching is used to adjust for selection bias in non-randomized studies to compare the effectiveness of interventions when there are significant baseline differences between the intervention groups. This presentation will discuss how to estimate the propensity score, form matched sets of subjects, assess the similarity of baseline factors between intervention groups, and estimate […]

Predictive Analytics Using Partial Least Squares Structural Equation Modeling

In recent years, Partial Least Squares Structural Equation Modeling (PLS-SEM) has emerged as a complementary SEM technique to analyze complex relationships among latent variables. Unlike traditional SEM, PLS-SEM is better suited for exploratory research where the goal is to best explain the observed data and/or make predictions about unobserved data (i.e., predictive analytics). Using a […]

Constrained Randomization: Applications to Group Randomized Trials

This seminar will include a discussion about different types of randomization strategies, focusing on applications to group randomized trials. Novel randomization strategies will be outlined, along with pros and cons of each. The use of one of these new strategies in the context of an intervention to improve colorectal cancer screening will be highlighted. Dr. […]

Time Series Analysis of 30-day Readmission Rates: Program Evaluation and Causality

This seminar will present a time series analysis of 30-day readmission rates using data from the BRIDGES project. The presentation will include a description of times series analysis and how it differs from other analyses of longitudinal data, the use of interrupted times series for program evaluation, how different times series can be cross-correlated to […]

The Use and Misuse of Statistics in Medical Research

The misuse of statistics in medical research is more common than one might think. While some may be due to use of the wrong statistical method, it more often reflects a lack of understanding of the purpose of statistical analysis, assumptions of inferential statistical methods, and the connection between statistical methods and research hypotheses. This […]

Assembling and Analyzing Cohorts from Electronic Health Records: An Example from the NICU

Electronic health records (EHR) contain a wealth of information potentially useful for public health research, and can be used to assemble retrospective cohorts for epidemiological analyses. This presentation demonstrates the experience of the Neonatology Division of the Department of Pediatrics to create a research dataset from the neonatal intensive care unit (NICU) EHR. We then […]