RSS 2.0 Feed

A & S News

Statistics: Devising Trials With Fewer Errors

May 26, 2009

Statistics: Devising Trials With Fewer Errors

The excerpt below is taken from "Number Sense and Sensibility: Math in a Modern World" by Leslie J. Thompson and originally appeared in the spring issue of the Arts and Sciences Magazine. The full text is available here.

Statistics: Devising Trials With Fewer Errors

A number of Baylor's statistics students are very involved in the world of pharmaceuticals, thanks in large part to a new partnership with drug manufacturer Eli Lilly that provides students the opportunity to work on challenging statistical research problems.

"We help define the model for clinical trials and try to provide the ability to determine how many patients are necessary to draw a valid conclusion about a certain drug," explains Dr. John Seaman, professor of Statistics. He is referring specifically to Bayesian inference, his graduate students' major area of research, which provides a means of accounting for uncertainty in things that cannot be seen in their entirety, such as a large population.

Think of it this way: If you're developing a new medication to treat high blood pressure, you can't test it on everybody in the U.S. who suffers from hypertension in the clinical trials. Instead, you work with a sample--a small group that is representative of the larger population. Seaman and his students not only develop the models to put together these groups, but also work on adaptive designs, which attempt to reduce the number of patients involved in a clinical trial while still providing accurate data.

"Our students have gone to Eli Lilly, the Mayo Clinic, the FDA--these places all engage in or monitor clinical trials," he notes proudly.