Stephen Bennett Johnson, PhD

Professor
Department of Biomedical Informatics
Columbia University

Research Focus

My research explores information systems that support collaboration in the biomedical domain. This work develops models of how biomedical professionals work together, in terms of specialized patterns of workflow, language use and social organization. Over the past twenty years, I have developed and tested these models in several areas of application, first patient care, then clinical research and now translational research.

Past Research

  • Flexible online medical record systems: clinicians need sophisticated online systems to to share information during the process of patient care. We developed a methodology for analyzing and representing the activities of medical professionals, and produced an abstract model of clinical events, participants and resources. We applied the model to create generic, flexible information systems that function in many settings, for multiple stakeholders, for long periods of time.
  • Expressive, language-based clinical systems: the language of health care professionals exhibits unique structures in syntax, semantics, morphology (structure within words, numbers and special symbols) and discourse (temporal, logical and persuasive aspects of language). We developed and applied models of language to extract information from free-form text, respond to user questions and aid in constructing new texts. This led to the development of new models of health information systems that provide a rich and expressive medium for communicating the facts of about a patient's care, rather than force unnatural, restrictive forms of data entry. Using network analysis, we showed that that large, mature terminology systems for health care resemble natural language more than they do systems of logic. This work points toward a new paradigm in which machines adapt more closely to clinician behavior rather than the other way around.
  • Reengineering clinical research: clinical research has not been well supported in the past by information technology. In particular, there is a lack of understanding about the social and behavioral factors that determine adoption and sustainability. We developed a framework for information technology adoption, and tested it in a clinical research network comprised of community practice settings. We conducted detailed observations of workflow (what clinical researchers actually do) rather than modeling protocols (what they are expected to do). Complex settings like clinical research can benefit from new kinds of information systems drawing on principles of social networking.

Current Research

  • Collaboration in translational science: I direct the biomedical informatics resource at the Irving Institute for Clinical and Translational Research. We are conducting research on how investigators establish collaborations, access services and share knowledge as they carry out clinical and translational research. We are developing a new system called WorkWeb, a simple, Web-based platform that enables investigators to search for collaborators, find expert assistance, place requests for services, and share materials using wiki technology. Another new system is called Sciologer, a Web-based, social network visualization tool that helps investigators to explore scientific communitities, discover groups, understand major topics of research and search for collaborators.
  • Collaboration in neuroscience: I am director of informatics for the Simons Foundation Autism Research Initiative (SFARI) . We are conducting research on informatics tools to support a community of investigators exploring the cause of autism, at genetic, synaptic and behavioral levels. We have developed a Web-based system called SFARI Base that enables scientists to access a large collection of phenotype and genotype information on families affected by autism. Investigators can place requests for DNA samples, explore 6000 phenotype variables, develop complex queries and visualize genetic variations in selected cohorts.

Opportunities for Students

  • WorkWeb: an online collaboration tool for clinical and translational investigators. The student can develop innovative social networking tools for scientists (need to know Perl). Students can also evaluate needs for, usage of or adoption of collaboration tools (need some background in social science research).
  • Sciologer:an online visualization tool to explore scientific communities. The student can develop innovative tools to search, visualize and navigate scientific communinties using data about publications and grants (need to know Java). Students can also evaluate needs for, usage of or adoption of visualization tools (need some background in cognitive science research).
  • SFARI Base: an online system for querying and visualizing phenotype and genotype data in autism research. The student can evaluate needs for, usage of or adoption of query and visualization tools in neuroscience (need some background in cognitive science research).
  • AQUA: the AcQUisitive Analyzer is a suite of tools with the goal of learning the structures of biomedical language with as little human input as possible. The student can develop discovery algorithms in syntax, semantics, tokenization or discourse (need to know Java, and have an interest or background in linguistics).

Contact Information

Address 622 West 168th St
  Vanderbilt Clinic, Fifth Floor, Room 558
  New York, NY 10032
Phone 212-305-1858
Fax 212-305-3302
Email My initials plus the numeral two, at this institution.