A software engineering framework for biomedical diagnostic systems

Abstract

Development of intelligent systems to support biomedical applications differs for traditional approaches to systems development. A large number of features needs to be extracted from data and processing of these is not satisfactory by conventional approaches and individuals. Development of such systems greatly changes the amount and nature of information available to physicians, and also the work involved in treating patients. Intelligent systems are learning-based and that makes them easier to adapt when diseases evolve or viruses mutate. This paper presents the use of an electronic nose and a neural network for classification of bacteria. It demonstrates how physicians can utilise it, in order to target their limited resources to specific patients. It also discusses how this work can be generalized in other similar domains and the lessons to be learnt.