Artificial neural network classification of UK hospitals using National Health Service indicators

This paper presents an investigation into the use of artificial neural networks for the analysis of health information. The Kohonen Self-Organising Map technique was used to group two data sets which were produced using the UK National Health Service Indicators: the first set contained indicators reflecting measures of composition and efficiency; the second set contained indicators reflecting measures of composition, efficiency and quality. Results obtained (which were subsequently confirmed by the use of statistical tests) differed significantly from those obtained using traditional hospital grouping techniques, suggesting both that the neural network approach to classifying hospitals is valid and useful and that existing perceptions of membership of UK hospital groups may be challenged.