A Field with Many Paths In and Out
By Rita Kukafka, DrPH,
Associate Professor of Biomedical Informatics and Sociomedical Science
Biomedical informatics is the study of information and computation in biology and health. We use the word biomedical in our name, the Department of Biomedical Informatics, to refer to our application area, but we mean the broadest interpretation, including all aspects of biology, health care, prevention, and public health.
Department researchers study and manage biomedical information, study behavior related to decisions, develop computational methods, use computational methods to generate knowledge, and use information and knowledge to influence behavior.
The 30 faculty members and 60 students work in a highly collaborative and interdisciplinary environment in a broad range of areas, applying biomedical informatics methods on scales from atoms to world populations. Using social network analysis, we found that our researchers fall roughly into three communities of practice, illustrated in the figure on the next page.
Empirical discovery and prediction (blue) is the computational simulation of biological systems focused on genomics and proteomics, data mining and knowledge discovery, machine learning and intelligent systems, and visual simulation of human biological systems.
Human and organizational factors (green) include decision support, health technology evaluation, consumer health, electronic health records, public health systems, cognition, human computer interface and usability, and the organizational impact of information and technology.
Information management (red) represents clinical information systems, safety and quality improvement, knowledge management, information standards and interoperability, telemedicine, concept representation, and natural language processing.
Many of our methods have been drawn from other fields computer science, management and decision science, biostatistics, engineering and information technology, cognitive and social science, operations research, physics, and applied mathematics and others have been developed within our evolving discipline. Many of our graduate students are drawn from other fields. The backgrounds and research interests of our students reflect the inter-, multi-, and translational nature of biomedical informatics and the highly collaborative work of our department.
I asked three graduate students to explain their path to biomedical informatics.
Daniel Stein, M.D.
Since high school I’ve been tinkering with technology and dreaming of how it could change lives. As an undergraduate biology major, I studied how computers could directly interface with our brains and how they could better the lives of patients with debilitating diseases. In medical school I conducted clinical research, evaluating diagnostic imaging techniques in the field of ophthalmology. These projects showed me how technology can improve our ability to detect and treat disease and reduce the human error intrinsic to the practice of medicine. They inspired me to seek more training in a field that combined my interests in medicine and technology and that would more formally introduce me to interdisciplinary research methods. My Ph.D. work focuses on collaborative technologies and how they can be used to increase healthcare quality, to prevent medical errors, and to provide clinicians with critical data to support patient care. I am particularly interested in care transitions, a precarious moment in patient care that could benefit from technological innovation to improve safety and quality. I am currently studying and developing tools to support safer periods of inpatient “cross coverage” the time when a hospitalized patient is cared for by a covering physician as opposed to the primary doctor or team.
My research is about processing biomedical text, such as clinical narrative and PubMed literature, also known as biomedical natural language processing (BioNLP) to reduce people’s cognitive overhead of reading and digesting enormous amounts of textual information in biomedical research and practice. Applications of BioNLP include automatic extracting/encoding of clinical reports, using electronic patient records for biosurveillance, and generating hypotheses via synthesis of scientific articles. My dissertation focuses on designing novel methods of using existing biomedical knowledge bases to improve the infrastructure underlying development of general BioNLP applications. After taking a few electives related to biomedicine in my senior year of college, I started to explore the potential of applying information science within biomedicine. I had to strengthen my background in biomedicine, computer science, and even math and statistics. One fascinating nature of biomedical informatics is that it allows you to pursue a wide spectrum of research, from extremely applied to extremely theoretical. Taking BioNLP as an example, you can develop information extraction systems serving as part of clinicians’ daily workflow, or you can study linguistic properties of biological sequences or even networks. Translational research also offers abundant opportunities, and we want to demonstrate the value of biomedical informatics in facilitating such translation. I am glad I have found this playground full of challenges and uncertainty and, more importantly, I know I participate in defining a growing discipline in history.
Jessica S. Ancker
I’ve never been able to decide whether I am a word person who likes numbers or a numbers person who likes language. My current research focuses on ways to explain numerical information (such as medical risks) to patients who might not have much background in math. After working as a journalist, writer, and editor, I took a job running a hospital department that provided manuscript editing for doctors. My mentor there encouraged me to learn some statistics to better understand the research methods and data in the manuscripts. The experience reminded me how much I enjoyed math, and I came to Columbia’s Mailman School of Public Health to earn my master’s degree in biostatistics. I found that my years as a writer gave me a unique perspective into explaining statistical concepts to non-statisticians. As a Ph.D. candidate in biomedical informatics, my dissertation work has been a series of qualitative and quantitative studies of different graphics to illustrate health risk information. The work integrates behavioral theories of risk communication with cognitive and perceptual research on decision-making, processing of visual images, and computer interaction. The biomedical informatics department is ideal for exploring multidisciplinary issues, bringing a wide variety of perspectives to bear on the study of biomedical information and the best ways to help people use it.