SIGNAL-BASED ANALYSIS OF CLINICAL MICROBIOLOGY

Hospital infections and antimicrobial resistance are scourges of modern society, affecting several million Americans annually and wasting billions of dollars. Timely, accurate data analysis is critical to preventing the spread of infection and resistance. Preliminary results show that treating clinical microbiology data as signals and applying various analytical techniques to them has tremendous potential to improve the detection of outbreaks and shifts in resistance early on, while the possibility for effective intervention still exists. Building on these results, this research aims to fully characterize at least five years of historical microbiology data from three prominent hospitals. This will be used as training data to optimize early detection techniques through the use of genetic algorithms (GAs) using a Linux supercomputing cluster. Surveys will be conducted of the infection control personnel before and after a one year trial of an early prototype real-time monitoring system to determine the impact such a system has on their ability to prevent the spread of infectious disease. Success will be measured by the survey results as well as by the number of outbreaks detected in a timely manner. This research will extend the scientific understanding of GAs, infectious disease surveillance, and cluster detection.