[Colloquium] Seminar Announcement: Data Mining for Network Discovery in Biology and Medicine - TODAY!
Ninfa Mayorga
ninfa at ci.uchicago.edu
Fri Jul 8 10:12:50 CDT 2011
Computation Institute Presentation
Speaker: KP Unnikrishnan, Ph.D., Research Assistant Professor of
Bioinformatics, Center for Computational Medicine and Bioinformatics
Host: Ian Foster
Date: July 8, 2011
Time: 2:00 PM - 3:00 PM
Location: University of Chicago, Searle 240A, 5735 S. Ellis Avenue
Data Mining for Network Discovery in Biology and Medicine
Network analysis allows a deeper understanding of systems ranging from
health & disease to power grids. Analysis of large time-evolving data
sets from these systems poses unique data mining challenges. We
present temporal data mining methods and associated statistical
significance tests for analysis of two such systems: i) neuronal
networks in the brain, and ii) hospital networks in the US. We uncover
i) the functional connectivity (graphical structure) of neuronal
networks and ii) cascades of inefficient patient transfers in hospital
networks. We also demonstrate the use of these methods to track the
time-evolutions of these systems and the use of these methods to build
predictive models.
BIO:
Data Mining for Medicine: I develop data mining algorithms for
analysis of clinical and biological data. We mine Electronic Health
Records (EHRs) and Billing Records, along with Genomic, Proteomic, and
Metabolomics data using these algorithms to help create "personalized
medicine". We have established close collaborations with Computer
Scientists, Statisticians, Mathematicians, Bioinformaticians, and
practitioners of Clinical Medicine for this purpose. For example, we
have started the following collaborations at the University of
Michigan. With Prof. Theodore Iwashyna from Critical Care, we are
analyzing the network of ICU transfers across United States. This has
revealed cascades of inefficient transfers in the system. We plan to
eventually propose a pilot program for the State of Michigan to create
an efficient network of clinical transfers. With Prof. John Younger,
we are analyzing time-stamped patient data from the Emergency
Department to predict sepsis. This project is c urrently funded as a
pilot by the Michigan Institute for Clinical and Health Research
(MICHR), an NIH CTSA Center. With Prof. Charles Burant, we are
analyzing the data (both clinical and biological) from a cohort of
nutrition and obesity volunteers to create predictive models of
outcome. This is currently funded as part of the NIH P30 grant to
Michigan Nutrition and Obesity Research Center (MNORC). We are also
working with researchers and developers at the National Center for
Integrative Biomedical Informatics (NCIBI) to create data mining tools
suitable for translational and clinical informatics.
Data Mining for Neuroscience: This research thrust follows a three-
year long basic research program I led at General Motors Research. My
collaborators in this area are Prof. P.S. Sastry (Electrical
Engineering, Indian Institute of Science), Prof. Vijay Nair
(Statistics, University of Michigan), and Prof. Anne Smith
(Anesthesiology, UC Davis). We believe that by studying complex
biological systems from a computational view point will lead to
organizational principles that not only reveal how the natural system
functions in its environment, but also suggest ways to engineer better
artifacts endowed with similar intelligent behavior. In the long term,
my research along this line will lead to better understanding of basic
mechanisms of perception, and memory and also lead to applications in
better neuro-prosthetic devices.
Network Analytics: Biological, other natural, and man-made systems
form networks at many levels. Throughout my career, I have been a
champion of interdisciplinary sciences. I have begun exploratory
collaborations with colleagues from LS&A, Engineering, Information,
Public Health, and Medicine at the University of Michigan in this
regard. I plan to create, co-ordinate, and participate in large inter-
disciplinary studies of networks, thereby helping create the field of
Network Analytics.
More information about the Colloquium
mailing list