SBIR-STTR Award

The goal of this research is to develop a next-generation software tool to assist the biostatistician in performing meta-analyses that combine summary information from multiple clinical or epidemiological studies. In Phase 1, we plan to collaborate with statisticians experienced in meta-analysis to select appropriate data management, quality evaluation statistics and graphics techniques from their own research and from the literature, and to combine these techniques into an initial prototype. Our experience in Phase I will form the basis for a Phase II proposal focused on completion evaluation, validation and refinement of the software. The major technical innovation in this project is expected to be the development of a new software framework for meta-analysis that supports a modern graphical user interface selection from the best available statistical combining and graphical presentation techniques, extension through inclusion of new algorithms, and automatic documentation of the meta-analysis process to facilitate review. The major health-related contribution is expected to be an increased ability for biostatisticians, clinicians, and epidemiologists to use meta-analysis as a tool to detect subtle responses and effects and to identify differences in the effectiveness of alternative therapies at an earlier stage of testing.Awardee's statement of the potential commercial applications of the research: The software to support meta-analysis of clinical and epidemiological studies has direct application to academic and Government biomedical research. The software also has application in clinical trials in the pharmaceutical industry.National Center for Research Resources (NCRR)

The approach known as "meta-analysis" is becoming increasingly relevant for clinical research and epidemiology because of requirements to obtain reliable quantitative estimates of relatively small risk factors and treatment effects. The goal of this project is to develop a new software tool for meta-analysis that incorporates the latest statistical techniques (including, for example, the hierarchical Bayesian models recommended by the NRC Report) along with graphical methods for user interaction, interpretation, and presentation.Phase I, based on a collaboration with the Statistics Department at Colorado State University, established key requirements for a capable and comprehensive meta-analysis tool, designed a framework and a set of innovative graphical and statistical techniques to meet the requirements and prototyped elements of the design for evaluation and critique. Phase II will be based on a broader collaboration with leading theoretical and applied researchers in meta-analysis and will aim at Beating a full-scale prototype for deployment and evaluation at a series of "beta sites".National Center for Research Resources (NCRR)