In the NNLM Big Data in Healthcare: Exploring Emerging Roles course, we asked participants, as they progressed through the course, to consider the following questions: Do you think health sciences librarians should get involved with big data in healthcare? Where should librarians get involved, if you think they should? If you think they should not, explain why. You may also combine a “should/should not” approach if you would like to argue both sides. NNLM will feature responses from different participants over the coming weeks.
Written by: Kristin Whitehair, Director of Library Services, St. Luke’s Health System
During the rise of evidence-based medicine, there was a clear link to the health sciences library. With evidence, usually as published in the literature, creating the foundation of practice, the library was a natural partner for clinicians, researchers, administrators, and students.
Now with the growth of big data and data science, we are seeing a similar transition. Organizations are devoting significant resources and energy to data science initiatives. The potential of data science appears huge, and largely untapped. The potential of data science has mostly focused on health research. However, data science can also look internally within the organization, especially for larger health systems. For example, retail corporations study online customer behavior, product offerings, and facility design. This same potential holds true for the health sector. Libraries can support data science in both health research and for organizations internally seeking to optimize business operations.
While there was a clear path to the library with evidence-based medicine, with data science librarians must build their own pathway. Part of this lies in how we define ourselves. Libraries can be a collection of literature, physical space, research expertise, and so much more. In general, libraries avoid limited definitions of their function. My library is a digital library, and I stress that we are a service with 24/7 access. This is an attempt to combat the stereotype of libraries as a room with books. By thoughtfully identifying our function and mission we can position libraries to take advantage of new opportunities such as supporting the organization’s data science initiatives, and whatever else may come next.
Additionally, libraries can provide resources to support data science initiatives. Some ideas that come to mind are coordinating coding boot camps and organizing regular interest group meetings. Throughout my career I’ve witnessed how the library can bring people with similar interests from different disciplines together. Public health researchers may be encountering the same technical problems as biostatisticians. The library can provide a forum for them to connect. All of these can be done by the library connecting people with similar interests.
Moreover, library staff can also develop knowledge and skills in the data science field. Broadly, there are two types of knowledge. First, there is definitional knowledge, to have an understanding of the meaning of terms. This is similar to a librarian having a broad understanding of cardiac terminology to better help cardiovascular researchers find information. Secondly, there is functional knowledge needed to perform data science tasks. This can focus on hands-on experience with data sets and popular data analysis programming languages. Over the course of the “Big Data In Healthcare” class we’ve seen several examples of both types of knowledge.
Building strong relationships throughout the organization is the key to creating services and developing skills that meet the organization’s needs. In general, library services are not “one size fits all.” It only makes sense that library services supporting data science would also not be. Strong organizational relationships are important to knowing what the key challenges and opportunities are for your organization, and are the key to ensuring that the library is best serving stakeholders.
In library efforts with data science, it is important to acknowledge where a library may not be a good fit. This depends on individual staff skills and attitudes. Much data science work is done using command line programming, which can be challenging to some. Personally, I have a strong grasp of descriptive statistics, but my knowledge of calculus is lacking. This creates a notable knowledge gap in the supporting data science. I need to know my limits in interpreting models. This is not a unique situation for libraries, as it is similar to when a library staff member is asked for medical or legal advice. We can provide information, but if lacking the appropriate qualifications should be careful when we offer an interpretation.
Overall, the growth in data science is an opportunity for health care in general, and health sciences libraries. We can all create our own path supporting these initiatives that is the best fit for our individual organizations.