2024 Joint Biostatistics Symposium

When: -

Where: 115 Biomedical Research Tower
460 W 12th Avenue
Columbus,  OH  43210

Register by March 31 to attend this free event.

Symposium Schedule

  • 11 a.m. - Registration and poster set up
  • 11:15 a.m. - Introduction
  • 11:20 a.m. to Noon - Mei-Cheng Wang, Department of Biostatistics, Bloomberg School of Public Health, Johns Hopkins University 
    "Cross-sectional data, epidemic dynamics and beyond"
  • Noon to 1:10 p.m. - Lunch and poster session
  • 1:10-1:50 p.m. - Ming Wang, Department of Population and Quantitative Health Sciences, School of Medicine, Case Western Reserve University 
    “Methods for integrating data to improve the estimation efficiency of primary data analysis”
  • 1:50-2:30 p.m. - Larisa Tereshchenko, Cleveland Clinic Lerner College of Medicine 
    “Applying artificial intelligence to ECG analysis: balancing between Hope and Hype”
  • 2:30-3 p.m. - Break and poster session
  • 3-4 p.m. - Fode Tounkara, Department of Biomedical Informatics, The Ohio State University 
    “The post-selection weighted elastic net approach for incorporating weak signals predictor in high-dimensional data analysis: application to HIV-1 drug resistance and kidney transplant studies”

Keynote Speaker

Mei Cheng Wang, Department of Biostatistics 
Bloomberg School of Public Health 
Johns Hopkins University

A cross-sectional population is defined as a population of living individuals at the sampling or observational time. Cross-sectionally sampled data with binary disease outcome are commonly analyzed in observational studies, frequently as an initial attempt, for the purpose of identifying how covariates or risk factors correlate with disease occurrences.

At Johns Hopkins University, cross-sectional data analyses using standard methods (testing statistics, logistic regression, et) are commonly conducted in doctoral dissertations by students with public health or medicine majors. Publications involving such data analysis can also be easily found online by searching the key words such as `logistic regression' or `logistic model' and `cross-sectional data' or `cross-sectional study.' It is generally understood that cross-sectional binary disease outcome is not as informative as longitudinally collected time-to-event data, but there is insufficient understanding as to whether bias can possibly exist in cross-sectional data and, if it exists, how the bias is related to the population risk of interest.

In this talk we study bias of absolute risk, relative risk and odds ratio arising from cross-sectional data, and connect the so-called "survival bias" to case-control data sampled from cross-sectional population. While the presence of bias may not be surprising, the bad news is that the bias is likely to change the interpretation toward the wrong direction. With auxiliary information of lifetime distribution, we present a bias-correction method which reassigns a portion of the observed binary outcome, 0 or 1, to the other disease category. Recommendations are discussed/invited at the end of this talk to find ways to provide advice to our project collaborators, students and colleagues.