Research in biostatistics covers a wide range of theory and methods applicable to biology and medicine. Some topics are stochastic modeling of biologic processes (e.g., carcinogenesis, cancer metastasis), modeling and analysis of medical images, pharmacokinetics and pharmacodynamics, statistical genetics, survival analysis, surrogate markers of clinical outcome, medical testing and screening (e.g., breast cancer screening), variable selection models, competing risks, missing data, measurement error, sequential methods for clinical trials, medical decision making, health policy, and statistical computation and software. These topics and others arise naturally from collaborative interactions among statisticians and biological and medical scientists.

The Department is home to the statistical office of a major NCI-sponsored cancer cooperative group, the Cancer and Leukemia Group B (CALGB). The PI (Dr. Stephen George) is a member of the Department and twelve members of the faculty serve as primary statisticians in various disease and modality research areas.

In the Duke Clinical Research Institute (DCRI), 15 members of the Department faculty are principal investigators or co-investigators for multiple NIH-sponsored clinical trials and outcomes research projects, and they have major leadership responsibilities in the numerous industry-sponsored trials conducted by the DCRI

The Medical Center has a strong history of collaborative research involving biostatisticians and clinicians. Recently, the increased focus on the quality of medical care and the use of quantitative evidence to support medical practice has enhanced these collaborations and generated an environment in which statisticians assume major leadership roles in addressing these important issues. For example, B&B Research Professor Dr. Jim Rochon is the PI for the Coordinating Center for the CALERIE study which is sponsored by the National Institute on Aging (NIA) of the NIH and is investigating the physiological effects of a diet that is severely restricted in calories. Other biostatistics research programs in progress include outcome studies of specific clinical treatments, studies of effective disease prevention, risk assessment for the development of diseases, clinical testing of procedures and drugs, and defining the most cost-effective therapies that preserve the quality of life. A major collaboration has begun within the DUMC to create a program to define and improve the quality of medical care using the Duke Health System as a model.

More scientists dually trained in the quantitative sciences and biology will be needed as the wealth of genetic information associated with the human genome project appears. The research areas include population genetics, genome informatics, genetic epidemiology, and statistical genetics. Duke already has a significant number of faculty members engaged in these activities and the B&B Department provides the necessary mathematical and statistical expertise.

One area of research of special interest is at the interface of molecular technology and population studies. Within the next few years, molecular genetic methods will provide the ability to genotype large numbers of people for a large number of DNA markers, including single nucleotide polymorphisms (SNPs) and raw DNA sequences. Statistical methods to use this information in genetic studies are needed. These methods will need to work in concert with rapidly developing bioinformatics tools. Evaluation of study designs and data collection methods will also be required as molecular genetic technology advances are incorporated into population studies of human disease and biological variation.

As the disease targets of human genetic study become more complex, the analysis of the data also becomes more complex. One of the pressing challenges in the analysis of the genetics of complex disease is the analysis of a large variety of demographic and clinical data describing the disease. Better methods to describe clinical epidemiologic variation with respect to genetic variation are needed. Development of statistical methods and population databases translating results from genetic studies in families and populations to individuals will be necessary if the promise of genetic medicine and pharmacogenetics is to be realized.

Statistical methodology is being developed at a rapid pace. However, practical evaluation and tools for application of these methods lag behind. In addition, there are existing statistical techniques, such as emerging methods and software for “mining” large databases, which should be evaluated in the context of genetic data. There is a pressing need to compile and evaluate existing and developing methodologies, especially in the context of optimal study design. The resulting study design information would allow development of a toolbox of complementary methods for analyzing rich epidemiological and genetic databases.

The Duke University Center for Computational Medicine (DUCCM) is an NIH Center for Immune Modeling for Biodefense and conducts research in the development and application of mathematical, statistical, informatic and computational methods for biomedical research, with emphasis on immunology, vaccinology and the study of inflammatory processes.

DUCCM operates both as a research group pursuing its own research agenda, collaborating with experimentalists who serve as co-investigators on these projects, and as a collaborative resource to other biomedical investigators on campus. It provides the core Biostatistics and Computational Biology support for the Duke Center for AIDS Research (CFAR), the Duke Radiation Countermeasures Center of Research Excellence (RadCCoRE), and the Southeast Regional Center for Excellence in Emerging Infections and Biodefense (SERCEB). Its own research agenda includes the development of microsimulation models of the immune response to vaccines, the development of a biomedical ontology for cell and molecular immunology, and the development of methods and software for multivariate biomedical assays.