New Leaders for New Times, Part 2

Robert L. Strawderman III, Sc.D.

Biostatistician Robert L. Strawderman III, Sc.D., is about to get his hands dirty. Harvard-trained, the former Cornell University scholar is no longer spending his days theorizing about study design and analysis. As the new chair of URMC's Biostatistics and Computational Biology department, Strawderman is also putting his visionary ideas to work. His shift in focus coincides with a revolutionary change in biostatistics: the emergence of big data.

"With the ever-expanding volume and complexity of the biomedical data becoming available, we are additionally being confronted with the challenge of helping researchers to formulate the relevant scientific questions," explains Strawderman. At its core, biostatistics is a science that focuses on quantifying the impact of uncertainty when drawing scientific conclusions from biomedical data. Once viewed as a backroom endeavor, it is increasingly moving to the forefront in all fields of biomedical study.

"Statistics has morphed into an unusually interdisciplinary and exciting field," Strawderman says.

Statisticians thrive on problem-solving. They confront unimaginably complex data, seeing opportunities to help others shape important advances in medicine. Just as one solution does not fit all health problems, complex data raise numerous questions that can't be addressed with simple, "off the shelf" statistical methods. To be effective, biostatisticians like Strawderman must be in tune with the art of science.

With that perspective, Strawderman joined URMC in July of 2012. He replaced interim chair David Oakes, Ph.D. Oakes led the department after the sudden death in 2008 of Andrei Yakovlev, M.D., Ph.D. Yakolev had initiated a major expansion, and the URMC search committee praised Strawderman's style as perfect for building on that effort.

"He brings the vision of a great leader, the experience of a great scholar, the patience of a great teacher, and the thoughtfulness of a true collaborator," says Robert G. Holloway, M.D., chair of the search committee and a URMC professor of Neurology and Public Health Sciences.

"Traditional methods of study design and statistical analysis are quickly being outpaced by the ability of modern technology to generate vast amounts of data. To remain relevant — a mission that must include successfully preparing the next generation — the department needs to enhance its ability to engage in cutting edge methodological research while maintaining the same level of collaboration for which we are already known throughout the University," Strawderman says. Making Sense of Big Data.

Overseeing how the department responds to the vast amounts of data generated by new technologies will be among his biggest challenges.

The field of statistics, developed in the 20th century as a scientific way to answer specific questions about specific populations, is relatively young. In less than a hundred years, it has literally turned on its side. Picture a simple database table, with a lot of rows listing the population subjects, and a few columns containing the data values associated with them. But as ongoing technological advances allow scientists to capture a comparatively gargantuan amount of information, the database table has become much wider than it is tall.

"We are increasingly confronted with the opposite situation: data involving tens, to hundreds, perhaps thousands of rows, and potentially more than one million columns," explains Strawderman. "In many cases, time-tested tools and basic methods of statistical design and analysis, just don't apply anymore."

With the advent of personalized medicine — where treatments can be tailored according to genomic, imaging, cyotmetric and health history from individual patients — serious challenges arise in designing studies and methods of statistical analysis capable of identifying the causal effect of genes on disease. Spurious associations between genes and disease can be easy to find, and lead to the wrong treatment targets. Therefore, the overriding goal for biostatisticians is to keep research conclusions objective and honest.

"I have never encountered a researcher that I felt was dishonest," Strawderman says. "But a researcher that's not being careful can easily end up with data and/or analyses that have significant problems. Put simply, our job is to make sure the study results are well-supported." Rich History, Future Strides One of the department's greatest achievements was converting a mathematical theory known as sequential design and applying it to clinical trials.

The late W. Jackson "Jack" Hall, Ph.D., started the groundbreaking work in sequential analysis decades ago, enabling Rochester researchers to study fewer patients and stop clinical trials earlier while still obtaining powerful results. Sequential design later became a national model due to its cost savings and ability to provide early insights about the viability of an experimental treatment.

Heading into the future, Strawderman anticipates recruiting faculty and doctoral students internationally in the area of big data, with a strategic focus on research areas important to genomics (e.g., bioinformatics), imaging, and early-stage trial design. He also plans to tap regional resources and step up recruitment activities at area colleges and universities, something that's not been done much in the past within the department.

"A lot of people don't realize that you can pursue a PhD in statistics or biostatistics at almost no cost to yourself," he says. Once you enter this critical area of science — the job prospects are terrific."

Strawderman's department is already loaded with more than 20 tenured or tenure-track faculty, two fully funded training grants, several research grants from the National Institutes of Health, and professional staff to support a statistical consulting service.

Rochester is a trendsetter in another way too, Strawderman says. Many peer biostatistics departments are housed in schools of public health and come with heavier teaching requirements. His home is in the School of Medicine and Dentistry. However, one of Strawderman's goals is to create stronger links to University of Rochester undergraduates in Statistics and Computer Science.

"Rochester is experiencing a trend that's occurring nationwide, the swelling of enrollments in many advanced undergraduate courses in statistics," Strawderman says. "We'd like to add undergraduate teaching to our portfolio of activities, which would be a significant contribution to the University and would really make us unique among our peers." Public Health Perspective Strawderman's training and expertise are based in delivering high-quality, efficient health care. He was a professor of Biological Statistics and Computational Biology and Statistical Science at Cornell University, and professor in the Department of Public Health at Weill-Cornell Medical College.

His longtime research interests in patient survival and outcomes, and related problems of evaluating quality of life and comparative and cost effectiveness, make him relevant in an atmosphere of reform. We're front and center in this environment," Strawderman says. If anyone can sell the attributes of biostatistics and computational biology, he can.

"I've had dinner with a couple of outstanding scientists here and we end up talking about how you make the jump from the formulas that we all learned in math and basic statistics — getting from point A to point B — to what we really do, which is very complex. It stumps a lot of people," he said. "I tell them that the right answers aren't always obvious — and that it's as much an art as science."