My name is Richard Boyce a postdoctoral associate at the
University of Pittsburgh's Department of Biomedical
Informatics. My primary research interest is biomedical
knowledge representation and especially how to better support
biomedical decision making when pertinent information is
missing, inaccurate or uncertain. You
can download my CV by clicking on this link
We are living in an exciting era where an advance in the basic
sciences can quickly lead to a dramatic improvement in outcomes
for patients. Exciting new opportunities for individualized care
are emerging as we discover the subtle variations in each
person's genetic makeup that can influence their course during
disease and their response to treatment. However, this is also
an era when investigators and clinicians are being overwhelmed
by the pace of scientific advance so that no one individual can
stay abreast of all of the current knowledge in their field of
study. My biomedical and health informatics training has given
me insight into theories and methods for dealing with the great
complexity of current biomedical research. These include methods
for data visualization, knowledge representation (modeling),
data sharing, and information retrieval.
Publications
Journal Publications
Boyce R, Collins C, Horn J, Kalet I. Computing with evidence part I: A drug-mechanism evidence taxonomy oriented toward confidence assignment. Journal of Biomedical Informatics. 2009. doi:10.1016/j.jbi.2009.05.001.
Boyce R, Collins C, Horn J, Kalet I. Computing with evidence part II: an evidential approach to predicting metabolic drug-drug interactions, Journal of Biomedical Informatics, 2009, doi:10.1016/j.jbi.2009.05.010
Boyce, R., Rose, T., Chilana. iCODEHOP: a new interactive program for designing COnsensus-DEgenerate Hybrid Oligonucleotide Primers (CODEHOPs) from multiply-aligned protein sequences. Nucleic Acids Research 2009 Web server edition. doi:10.1093/nar/gkp379.
J. P. Staheli, J. T. Ryan, A. G. Bruce, R. Boyce, and T. M. Rose. Consensus-degenerate hybrid oligonucleotide primers (CODEHOPs) for the detection of novel viruses in non-human primates. Methods, 49(1):32-41, 2009.
Boyce R., Collins C., Horn J., Kalet I., "Modeling Drug Mechanism Knowledge Using Evidence and Truth Maintenance," IEEE Transactions on Information Technology in Biomedicine. Volume 11(4) 2007, Page(s):386 - 397. Digital Object Identifier 10.1109/TITB.2007.890842. see the abstract on IEEEExplore
Conference Publications
Boyce R., Collins C., Horn J., Kalet I., "Qualitative Pharmacokinetic
Modeling of Drugs," Proceedings of the 2005 America Medical Informatics Association conference. 71-75. PubMed ID: 16779004.
Boyce, R. Chase, R. "Conflation System MBP, Proceedings of the
International Conference on Imaging Science, Systems, and Technology,"
CISST'03, v.1, 246 - 252, June, 2003.
Presentations
Boyce R., Karras B., Lober B., "Pre-processing to Improve the
Classification of Chief Complaint Data," 2005 National Syndromic Surveillance Conference. click for a PDF version available online.
Rossini A., Boyce R., and Webster E. "The Statistical Reality Engine,"
Presented at Directions for Statistical Computing (DSC) 2005.
Boyce, R. "VRGL / StatVizVR - An Toolkit and an Environment for
Statistical Visualization," Presented at the National Library of
Medicine Training Directors' Meeting 2004
Boyce, R. "An Algorithm
for Automatic Multi-Source Image Conflation Using Linear Features,"
Presented at SOURCE 2003, Central Washington University and the
University of Washington Spring Research Conference, May 2002
Boyce, R. "The Synchronous Motor Project," 1994 McNair Scholars
Program Journal, Summer Research Abstracts
|
|
Current Research Interests
- Knowledge-based approaches to drug-drug interaction prediction and identification
- integrating emerging genetic and genomic knowledge into drug therapy decision support
- computational methods for simplifying biomedical knowledge-base development, curation, and use
- applications of truth and belief maintenance systems to biomedical knowledge-representation
Current Projects
Knowledge-based Approaches to Drug Safety
The focus of this research is how to best represent drug mechanism
knowledge for the purpose of predicting clinically relevant
drug-drug interactions. My contributions include a novel qualitative
model of drug-drug interactions and an evidence-centric knowledge
collection system that links changes in assertions about drug
properties to the set of predicted drug-drug interactions (Boyce
2009 Parts I and II, Boyce 2007). I am currently working on a
project that uses these methods and observational study techniques
to examine if the risk of experiencing falls increases for nursing
home residents taking drug pairs that are predicted interact by
metabolic inhibition. This project is supported by the University of
Pittsburgh Institute on Aging.
iCODEHOP - the Interactive CODEHOPs design program
I designed and implemented a tool called iCODEHOP that assists in
the design of degenerate PCR primers to detect novel proteins. The
aim of the tool is to help researchers in many areas of clinical
microbiology design effective DNA based diagnostics for infectious
organisms with less effort. iCODEHOP, is a web application that
links together several new and existing bioinformatics tools into a
set of intuitive workflows. Users can quickly scan over an entire
set of degenerate primers produced by the program to assess their
relative quality and select individual degenerate primers for
further analysis. The program predicts annealing temperatures for
degenerate primer pools, displays phylogenetic information for the
sequences covered by the primer, and allows the user to easily
design new degenerate primers from sub-selections of their input
sequences. The tool is currently hosted by the University of
Washington's Center for Public Health Informatics and can be seen
at the following URL: The
interactive program for creating COnsensus DEgenerate Hybrid
Oligonucleotide Primers
Previous Projects
Improving Signals used for Public Health Surveillance
An analogy for a target of Public Health Surveillance is that of a
signal relaying community disease state. In practice, PH officials may
monitor multiple syndromic and traditional signals to such as pharmacy
orders, school absenteeism, water treatment plant turbidity logs,
clinical diagnosis, surveys, and laboratory-test results. A syndromic
signal differs from a more traditional PH surveillance signal (e.g
case reports) in both the reliability and timeliness of the signal. As
part of ongoing research on methods for using emergency department
chief complaints as syndromic surveillance information signals, I
developed an automated method for normalizing free-text chief
complaints and investigated its its effect on the accuracy of chief
complaint classification of a naive Bayesian classifier, COCO. (Boyce 2005)
Collaborative Statistical Visualization - enhancements to statistical
consulting using the virtual world paradigm
The focus of my Master's degree was on building a platform that
enables a statistician and a biomedical domain expert to
collaboratively visualize multivariate bio-statistical data. We call
the platform the Statistical Reality Engine (SRE), it is an
environment for collaborative statistical visualization employing
Virtual Reality. The system is designed so that many views can be
rapidly prototyped and tested for usefulness.
The SRE is an R package that provides a platform for experimentation
with ideas for data visualization in Virtual Reality. How did we get a
complex multi-threaded platform like OpenSG and VRJuggler to work with
a interactive, single threaded, interpreter like R? The trick was to
use Duncan Temple Lang's experimental REventLoop
Package. The result is that VRJuggler's event loop replaces R's
and provides an interactive R session to users using Readline.
|