Department of Biostatistics

People

Prospective Students

Academics

Research

News & Events

Calendar

Consulting

Employment Opportunities

Resource Quicklinks

Computing Environment

Contact

THESIS DEFENSE ABSTRACT

Robust Statistical Methods for the Study of Disease through Complex Structural Outcomes

Russell Shinohara, PhD Candidate, Johns Hopkins Department of Biostatistics

In modern biomedical research, the quantitative sciences play a key role in our understanding of disease and the improvement of human health. In particular, the development and appropriate use of statistical methods are a cornerstone of public health research and practice. During my time as a graduate student at Johns Hopkins, my contributions have centered around two areas of biostatistical research: the first such area involves robust inference. In many public health contexts, decisions are made on the basis of statistical findings that depend on assumptions. These assumptions can describe our understanding of the structure of biological processes, the relationships between key variables of interest, and the performance of estimation techniques. In this work, we consider the robustness of statistical inference with respect to assumptions in each of these three genres. In the remainder of the work, we consider the analysis of data with complex structural outcomes. As our understanding of simple data structures is refined, the data we observe in modern applications is becoming more complex. Although the statistical analysis of survival data is now commonplace, the analysis of more complicated event processes often involves ad hoc techniques that are not well-suited for the questions of interest. We study population dynamics and estimation in such a scenario that is common in chronic disease settings. In addition, the analysis of high-dimensional data remains an active area of statistical research, and we study the analysis of imaging outcomes for improving our understanding of etiology and for the development of clinical biomarkers.