Data science is critical to driving innovation in healthcare, but it requires planning, communication and multi-level coordination for big data to be available, protected and usable in a way that can accelerate discoveries.

That means building new tools for data analysis, growing a workforce of data-savvy scientists, researchers and clinicians and shifting the thinking about science away from domains to more of a social value - work that is already underway across the National Institutes of Health, said Patricia Brennan, director of the National Library of Medicine, speaking Sept. 20 to University of Delaware students and faculty attending the Health and Big Data Forum.

“We recognize the need to build for a future, not just build for what we have now,” said Brennan, who also serves as the NIH interim associate director for data science. “We need to work with our academic colleagues to change the culture of discovery and to add to epidemiological experiments and observational research the ability to also grow science out of data.”

Brennan, a UD alumna, is the first woman and the first nurse to serve as head of the library since its founding in 1836. She joined a panel of Delaware experts to discuss the opportunities and challenges that big data offers and also share how data science is being used to propel research in the state. The UD Research Office and Delaware INBRE hosted the event.

Big data offers great potential for improving and transforming our world through the capture, mining and analysis of data. Still, there are myriad challenges in securing access, sharing information, determining how long data should be stored and even figuring out what to do with data that doesn’t look like numbers, such as motion capture.

“Data science is not just statistics on steroids,” Brennan said.

At the state government level, there’s still a great need for data science to help inform decision-making, particularly relative to those with the most pressing needs in Delaware, said Rita Landgraf, director of the UD Partnership for Healthy Communities. The state took a big step in 2016 when then-Gov. Jack Markell launched an Open Data Portal that provides non-identifiable public data, but even more data needs to be made available and used, Landgraf said.

Financial constraints and elected office terms can put the focus on short-term problems without focusing on long-term issues that could be better addressed by identifying and measuring key indicators, including social determinants of health.

“State decisions are being made with non-scientific data,” said Landgraf, also a Professor of Practice in the College of Health Sciences, who joined UD this year after serving eight years as Delaware Health and Social Services Secretary. “It’s how you use data, not just for today, but getting ahead of it.”

It’s an unfortunate reality that healthcare providers lack the research evidence needed to support the medical decisions patients and physicians have to make each day, said Dr. Vicky Funange, operational vice president for research Nemours/Alfred I. duPont Hospital for Children. Pediatric patients often have to rely on hand-me-down evidence from adult patients.

To help improve children’s health through research, Nemours partnered with seven other children’s hospitals to create PEDSnet, a learning healthcare network that conducts research and clinical trials using data from electronic medical records.

“We wanted to use that big data, pulling information from all of those health records to generate knowledge that would improve patient and system outcomes,” she said.

The challenges of big data have less to do with the quantity of data than the collaboration that’s required, said Dr. Steven J. Stanhope, associate Vice President for Research and Director of Delaware INBRE. There can be issues with conflicting goals, identifying roles in multi-partner projects and understanding the implications of resource sharing.

As a result, it’s critical for institutions to identify, establish and support organizational factors that promote the effectiveness and productivity of team science, said Stanhope, who also directs the BADER Consortium. Those factors can include inter-institutional management and partnership models that encourage collaboration and goal congruence while reducing the strain on the team.  

“Big data fits squarely into the team science realm,” Stanhope said.

At UD, there’s been a deliberate effort to create an interdisciplinary, cross-campus approach to data science through the Center for Bioinformatics and Computational Biology (CBCB). Now in its ninth year, the center has 60 affiliated faculty from five colleges as well as external partners including Christiana Care Health System, Nemours/Alfred I. duPont Hospital for Children and Delaware BioScience Association, said Dr. Cathy Wu, Edward G. Jefferson Chair and CBCB director.  

The CBCB is actively engaged in collaborations at the state, national and international level and also is part of the NIH Big Data to Knowledge (BD2K) initiative, started in 2012 to help with biomedical research discovery and increase engagement.

Wu said an early decision for the CBCB was the establishment of graduate and educational programs in bioinformatics and computational biology. “For this kind of multi-disciplinary activity, students really serve as the best bridge to connect researchers that would otherwise not be thinking about collaborating together,” she added. “Students can be co-mentored and we can use that as a bridge to bring investigators together.”

For Brennan, it was UD that set her on the path that eventually led to her data-driven role at the National Library of Medicine, the world’s largest biomedical library.

“Walking out of the School of Nursing building on the other side of Main Street, and this article caught my eye. It said, ‘What are computers doing in health care for the lucky few nurses who can use them?’ ” Brennan said. “What we know now is that technology drives innovation and we need that more than ever in healthcare.”