Useful statistical ideas are drawn into biomedical practice from a wide range of sources, including theoretical research and developments in other application areas. The long-term purpose of this grant is to speed the transfer of promising new statistical technology into health- related applications.
Five specific aims are proposed here, all of which involve a combination of computer-intensive methodology with new developments in statistical theory. Bootstrap and permutation methods are suggested for some specific biomedical problems: in the analysis of accuracy measurements for medical images, and also for chromosomal ordering via radiation hybrids. More general problems of the types encountered in biostatistical work are also under investigation: setting accurate confidence intervals, assessing the accuracy of discrete estimates such as phylogenetic trees, and making use of Bayesian ideas in an objective manner. The investigation pursues these topics from several vantage points, the Stanford statistics department, the medical school, and the Human Genome project.
Showing the most recent 10 out of 11 publications