Bioinformatics and Machine Learning

The committee felt that setting up a unit in direct competition with other groups and advertising ourselves as a full resource is not a good idea. Rather, we should lobby biomedical research leaders regarding our added value (see below). We should accelerate recruiting of MS biostatisticians with general knowledge of genomics and proteomics, but hold off for now in recruiting PhD faculty specializing in these areas.

Department members working in bioinformatics and other areas related to machine learning need more cohesion. They need to meet more often. The department also needs to be more visable to the School. Some ways for doing this are
  • Have a monthly GCRC-type lunch where methodology and problem datasets are discussed
  • Attend grand rounds, departmental research conferences, and other important conferences run by biomedical researchers, and take part in the discussion
  • Have our faculty give "big talks" to other departments
  • Be visable at Cancer Center retreats

Added Value of Biostatistics

  • Biostatisticians understand the measurements, which leads to be inference regarding the question of interest
  • We are collaborators from the beginning and can help prevent major design problems from ruining the research
  • We are good at choosing analysis procedures that fit the biomedical problem
  • All of this leads to improved repeatibility of research results
  • We are major players in the QC of science and the science of science
  • One of our main jobs is to protect the integrity of science using principles developed over the past 4 centuries

Specific Tasks

  1. Spend more effort doing simulation studies to document the performance of methods we and machine learning experts use
  2. Routinely supplement analyses from ourselves and others in which the outcome data are randomly permuted so that responses are associated with the wrong patients; this is a great educational tool when apparently valuable results may be obtained when we are sure no associates can exist
  3. Spend more effort teaching state of the art data visualization techniques and using these in our everyday practice
  4. Collect data and examples:
    • "before and after" where you have helped to re-write specific aims in grant proposals
    • how you benefitted research during a consultation, especially regarding questions or input from you that resulted in a change in experimental design or in how measurements were made or transformed
    • consider working with Rogers Hall in the Dept. of Teaching and Learning at Peabody who is conducting research on statistical consultation

Topic revision: r2 - 26 Apr 2013, JohnBock

This site is powered by FoswikiCopyright © 2013-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback