Biostatistics is a statistical discipline focusing on theory and application of statistical methods for the analysis of problems related to biology, medicine and public health. The ultimate intent of Biostatistics is to better understand the factors that affect human health through judicious use of statistical methods.
The colossal advances in computational methods in the last couple of decades transformed Bayesian Inference from somewhat of a conceptual statistical paradigm based on philosophical consistency to an integral statistical tool for analyzing today’s highly complex data. Powerful computational tools allow Bayesian methods to tackle large and complex statistical problems with relative ease where frequentist methods can only approximate or fail altogether. Our faculty are involved in developing modern Bayesian methods for change point analysis, clinical trial designs, correlated time series processes, functional data analysis, genetics and genomics, network meta analysis.
Big Data refers to data that are not only “Big” in their size, but also too complex to be analyzed using standard analytic tools – they are indeed challenging in many ways. Biomedical data from research laboratories that use advanced biomedical technology such as microarray, next generation sequencing and DNA Methylation Beadchips are some examples of “big” biomedical data. Our faculty conduct research of Big Biomedical Data modeling, with specializations in (i) Computational Genomics by using computational tools to analyze large scale high-throughput genomic data; (ii) Epigenomics by developing methods to understand cell modifications that are indicative of complex diseases; (iii) Population Genetics by using DNA sequences sampled from populations to study population size changes, migration and admixture; (iv) Association and Linkage by developing new statistical methods and computational algorithms to study the genetic basis of diseases; and (v) Data Mining where exploratory data analysis is performed for generating biological/medical hypotheses.
Change point analysis is a collection of statistical methods concerning the changes in the distributions or in the parameters of a distribution that describes the underlying model of any data naturally ordered either in time or in position. It has a wide spectrum of applications in industrial quality management, climatology, economics and finance, medicine, genetics, etc. Our faculty have specialty in Bayesian Change Point Detection and Parametric Change Point Analysis with applications to genomic data.
Generalized linear models are widely used in various applications and they embrace a large class of statistical models. In this research area, our faculty are actively involved in research of cluster analysis of continuous and categorical outcomes, modeling longitudinal data of multiple outcomes that change over time simultaneously, and mixed effect models by modeling repeated measures over time where the response is continuous, nominal, or ordinal.
Nonparametric statistical inference is a branch of statistics in which no assumption about the form of distribution of the population under consideration is made. Because of unknown population distribution, classical method will not work as they should be and hence may lead to wrong results. Resampling techniques such as permutation tests, cross-validation, jackknife and bootstrap can be incorporated in developing efficient and practical nonparametric procedures, when parametric methods may not be feasible. Robust methods are extension of nonparametric procedures in the sense that it deviates from parametric form of the population with small deviation, but still produces reliable and relatively efficient estimates and test procedures.
Translational research is about translating innovative and advanced results from basic sciences research to new diagnosis and treatment regiments for the enhancement of human health and wellbeing. Our faculty are actively involved in developing and expanding methods used in translational research, which include biomarker data analysis, clinical trials, latent variable modeling, messy data analysis and survival analysis, among others.