Rice Statistics alumna analyzes data, communicates its story at MD Anderson

As an associate professor of biostatistics, Christine Peterson '13 turns statistics into insights for improving human health.

Christine Peterson turns statistics into insights for improving human health

“I get to learn new things all the time. My skill set is in statistics and biostatistics, and a big part of my job at MD Anderson is to get data from collaborators and analyze it, then communicate the results back to our collaborators,” said Christine Peterson, who earned her Ph.D. in statistics from Rice University in 2013.

She is an Associate Professor of Biostatistics at the University of Texas MD Anderson Cancer Center, where she primarily focuses on the analysis of data with many variables (multivariate data), including gene expression and microbiome data. Her department is famous for its work in clinical trial designs, and the quantity and quality of their studies on newly developed cancer treatments make the department both unique and strong in that area.

“I enjoy consulting – meeting with people to design experiments and clinical studies so that the study design best matches their hypothesis. We are asking and answering questions like, ‘What kind of intervention will be tested? What factors might influence treatment response? What are potential toxicities or side effects of the treatment?’ 

“Understanding the question you want to answer is a big aspect of statistics, and that requires clear communication. Often that means translating a non-statistical question into a statistical one, and then the reverse once the data has been collected. When the raw numbers are trying to tell a story, careful statistical analysis is required to translate the data into scientific understanding,” she said.

Her first exposure to telling the story behind the statistics came through a statistical programming course at Rice. Although she had majored in applied mathematics at Harvard and worked as a software design engineer for Microsoft for four years, Peterson said this was the first time she was programming not to get a numerical answer but to produce a report with figures, descriptions, and interpretations of the work she’d done.

“We could not just focus on the ‘computer skills.’ We had to take the results of our R script and produce graphics and figures. That class laid a solid foundation for me to begin communicating scientific data,” said Peterson.

“I also had an excellent Ph.D. advisor in Marina Vannucci. She was great about providing me one-on-one training and guidance for going about the process of scientific publication – preparing figures, editing, revising, and presenting my research. All of it was so valuable.”

With more than 6000 citations and over 75 research papers, Peterson proves Vannucci trained her well. Since 2018, Peterson has typically published 10-15 papers each year. The pandemic did not slow her down; in 2020, she published 25 papers. 

“In addition to Marina, I worked with Tracy Volz and the group now known as the Activate Engineering Communication Program to get coaching and feedback on my presentation skills,” said Peterson. “Much of what I do now is talking with people in meetings, but a fair amount of my communication is in presentations at seminars and conferences. Learning to formulate and give those talks links back to working with Tracy; I even got her feedback on my Ph.D. defense.

“Ironically, I still have the same challenge today: condensing years or months of work down into a 45-minute talk. Distilling it down is hard; I want to capture those details I worked on for so long. But the audience does not want to hear all those details, so I use a higher vantage point and condense the message. Understanding how to do that is still a very valuable skill. I often practice on my family members to get feedback from someone outside my area of expertise.”

She believes a common mistake presenters make is to practice with an audience full of colleagues in their field. Today, there are often biostatisticians in her final audience but they do not necessarily work in the same area. Biostatisticians specializing in survival analysis or machine learning won’t necessarily understand the details of a genomics application.

To guide her audience through a talk, Peterson incorporates what she calls signposts. These pauses to recap a basic point help each person get something out of the talk and remain oriented in the talk’s progress. Successfully bringing her audience through the story is important to Peterson.

“Part of what I enjoy most about statistics is how it gives me a lot of flexibility to engage with different disciplines,” she said. “People are grappling with data that is hard to interpret and visualize. We’re looking at multivariate data with many features and trying to determine what is relevant to the outcome and how to present the results. That is fun for me!

“We’re also working on quantitatively challenging projects in omics, genomics, microbiology, and imaging data – which is really big data. These data sets can provide answers to questions and problems people are interested in solving. It is exciting for me to dive into messy data and find a way to tell a story about that data. When it comes together, it is so rewarding to see a new solution for cancer diagnosis or treatment evolve and to be a part of that.”

The thrill of success does not dim Peterson’s awareness that many of the studies she helps design enroll real people as subjects. That is one of the biggest differences between her Ph.D. research and her MD Anderson work. She said it is more critical now to think ahead and to both make and follow a statistical analysis plan. 

“These are human subjects. You cannot say at the end of a trial, ‘Oh, we should have designed that differently.’ From the start, we look at what is being measured and consider how we will define the study outcomes. Working with big data — like genomics data — we must understand how to focus our hypotheses before we begin the study. We must plan at the most rigorous level possible before we enroll human subjects, to ensure the study will yield clear, unbiased results.”

These opportunities for collaborative science drew Peterson to be interested in MD Anderson’s Biostatistics Department even before she applied to Rice. After talking with faculty in that department, she submitted her application to the Rice Statistics Ph.D. program, then returned to MD Anderson following her postdoctoral research fellowship at Stanford.

“Now I help people design their studies,” she said. “But I also get to supervise grad students and postdocs, so I am helping train up the next generation of biostatisticians. And I hope to continue Tracy’s and Marina’s legacy of shaping them into stronger scientific communicators.”