Introduction

There is a continuing need for the rapid and accurate identification of microorganisms, particularly in the clinical laboratory. Recent advances in analytical instruments have allowed the characterization of microbes from their phenotypic make-up, but these techniques tend to produce vast amounts of multivariate data that can be extremely hard to interpret. There is therefore a need to exploit modern statistical and related (chemometric) methods to facilitate automatic microbial identification. A particularly powerful set of methods is based on the use of artificial neural networks (ANNs). Over the last few years the availability of powerful desktop computers in conjunction with the development of several user-friendly packages that can simulate such ANNs has led to these ‘intelligent systems’ increasingly been adopted by the microbial taxonomist for pattern recognition. The nature, properties and exploitation of ANNs for the classification and the identification of microorganisms by whole-organism fingerprinting is reviewed.

In just about every area of microbiology the more rapid, but still accurate, characterization of microorganisms is a desirable objective. In medicine, shortening the time taken to identify a pathogenic bacterium, yeast or fungus will accelerate targeted prescription and should lead to improvements in epidemiological studies. In industry, speedy characterization will allow for better quality control procedures on both raw materials and finished products, and allow accurate microbial screening for isolates producing novel pharmacophores, thus saving time and money.