Rapid detection of bacterial infection using gas phase time series analysis

Abstract

Quick identification of bacterial growth in culture is demonstrated using time series analysis of gas phase data collected by a commercial gas sensor. Clinically standard blood agar plates were inoculated with different concentrations of Escherichia coli and incubated at 37 C. BME688 gas sensors and read-out electronics were used to measure the gases emitted during bacterial growth. The time series show sigmoidal features, which appear at earlier time points with higher initial concen- trations. A modified logistic model was found to fit the data well and confirms faster growth with higher concentrations. A data analysis algorithm was developed to identify the earliest time- point for bacterial growth detection; it revealed a logarithmic relationship between the initial seeding concentration and time for signal peaks. Concentration-dependent bacterial growth was identified at 2-6 hours, more than 90% faster detection, compared to the current clinical protocols that take up to 72 hours.