Neural network analysis in predicting 2-year survival in elderly people: anew statistical-mathematical approach

We designed this study to test the usefulness of artificial neural networks (ANN) in assessing 2-year survival in elderly persons, and to understand t he net's logical functioning, thus determining the relative importance of t he single biological and clinical variables which influence survival. ANN a re statistical-mathematical tools able to determine the existence of a corr elation between series of data and, once 'trained', to predict output data given input data. Although ANN have been applied in various areas of medica l research, they have only very recently been applied in geriatrics (Caccia festa et al. 2000. Arch. Gerontol. Geriatr. 31 tin press)). We built up an ANN to investigate how 17 clinical variables relating to a sample of 159 el derly people affect survival, and the possibility of predicting 2-year surv ival or non-survival for each single subject. When tested on a sample of 20 elderly people, the trained network gave the correct answer in 85% of the cases. We then extracted the mathematical function that the net used fur ca lculating the output (survival) for each set of input data (clinical variab les). Using this formula, we investigated how some clinical variables influ ence 2-year survival: we found that a low serum cholesterol level is an unf avourable characteristic in relation to survival. We conclude despite the f act that the sample studied was relatively small - that ANN are useful in p redicting 2-year survival in elderly people. The mathematical function we o btained from the not seems useful ill determining the relative importance o f single variables related to survival. (C)2001 Elsevier Science Ireland Lt d. All rights reserved.