Nosocomial infections cause about 90,000 deaths annually in the U.S. and have an associated medical care cost of about 3.5 billion dollars. Despite being the fourth leading cause of death, there has been limited development of rapid, integrated tools for determination of outbreaks of hospital-acquired infections. The goal of the proposed research is to test feasibility of development of software algorithms for identifying clusters of bacteria involved in nosocomial infections. This will be accomplished by creation of new algorithms for clustering bacterial fatty acid composition data to detect infection clusters and through the creation of a """"""""data mining"""""""" algorithm to provide patient demographic information needed to distinguish nosocomial outbreaks from community-acquired infections or pseudo-outbreaks. These software algorithms will be integrated into the MIDI Sherlock Microbial Identification System as a fully automated real-time epidemiology tool. Hospital infection-control personnel will be able to use the output to immediately implement infection control measures, and thus to reduce the impact of nosocomial infections.
NOT AVAILABLE
