How can we use big data to save cancer patients’ lives?
October 15, 2017
Stephanie B. Wheeler, PhD
Associate professor of health policy and management
Yes, big data can save lives.
My work uses complex computer programs to simulate how public health interventions and policies are likely to play out in reality. Think of it as “Sim City” (tinyurl.com/wiki-sim-c) for fighting cancer. I start by asking key questions, such as Which intervention is most cost-effective? and Which policy results in the best health outcomes over time? Then I build a realistic synthetic population and run different mathematical models to see how that population will respond.
My goal in these simulations is to inform public health policy and decision making. I want to see decision makers directing money and energy toward the courses of action that are most likely to be successful and valuable in the long run.
Q: How do simulations work? What are their pros and cons?
A: Simulations start with surveillance. To build a program that examines cancer, for example, we first have to identify the “hot spots” of cancer mortality, both geographically and within specific sub-populations, such as rural or minority groups.
We collect and analyze data from multiple sources, from cancer registries to census data to health insurance claims, and program those into the simulation to create a digital representation of a group of people. Once we identify a group of people with a disproportionate burden of cancer – for example, low-income people in rural North Carolina – we are able to overlay metrics from research studies that represent possible interventions and policies that could help address the gap in cancer outcomes.
The biggest “pro” of working with simulations is that they offer data-guided decision making and can be used to examine the role of uncertainty in our understanding. In addition to offering information on likely outcomes, they can reveal possible unintended consequences of an intervention. For example, when more people get screened for cancer and live longer as a result, what potential cost burden is added to the health-care system?
A primary “con” is that there’s still some resistance among medical providers and others to accept simulations as valid tools. Some people think of simulations as “glorified weather forecasting” – and, as my dad would say, the meteorologists always get it wrong! The reality is that we offer our conclusions with uncertainty analyses built in. Anyone working in simulation modeling has to be able to embrace uncertainty.
Q: How can simulations be used to reduce health disparities?
A: Public health has been characterizing health disparities for five or six decades now, and yet, very little meaningful progress has been made in reducing those disparities. My work is an attempt to follow through on those findings and proactively address health disparities.
While my current research focuses on cancer prevention and treatment, simulations can be built to model all kinds of public health concerns. Partnerships are essential to this work. The future of public health, in my opinion, is about changing how we work. Epidemiologists, health services researchers, economists, engineers, clinicians and behavioral scientists all need to sit at the same table together; that’s what we do in my research teams. I have learned more from my colleagues than I ever could have learned on my own or siloed within my discipline.
We need to make the leap from tackling one little piece of a large problem in isolation to working in interdisciplinary teams with leaders who can translate between fields – that’s the path to meaningful change in public health policy and outcomes.
Wheeler is recipient of UNC’s 2017 Philip and Ruth Hettleman Prize for Artistic and Scholarly Achievement by Young Faculty.
Carolina Public Health is a publication of the University of North Carolina at Chapel Hill Gillings School of Global Public Health. To view previous issues, please visit sph.unc.edu/cph.
