Biological agent detection and identification using pattern recognition

Abstract

This paper discusses a novel approach for the automatic identification of biological agents. The essence of the approach is a combination of gene expression, microarray-based sensing, information fusion, machine learning and pattern recognition. Integration of these elements is a distinguishing aspect of the approach, leading to a number of significant advantages. Amongst them are the applicability to various agent types including bacteria, viruses, toxins, and others, ability to operate without the knowledge of a pathogen's genome sequence and without the need for bioagent-specific materials or reagents, and a high level of extensibility. Furthermore, the approach allows detection of uncataloged agents, including emerging pathogens. The approach offers a promising avenue for automatic identification of biological agents for applications such as medical diagnostics, bioforensics, and biodefense.

Information fusion for CB defense applications: Challenge and opportunity

“…Performance studies for the above approach included identification of four pathogens including bacterial, viral, and toxin. The results of the computational experiments with the approach we developed [5] [6] indicated that, within the constraints of the data we had in that effort, high levels of identification accuracy can be obtained using that fusion approach.…”

“…), operation without the knowledge of pathogen's genome sequence, and the potential for identification of uncataloged, i.e., emerging pathogens. The biological genomic aspects of this effort are beyond the scope of this paper and can be found in [6]. The brief sketch that follows touches only on the aspects pertinent to the information fusion topic.…”