Dealing with missing values in neural network-based diagnostic systems
- 278 Downloads
Backpropagation neural networks have been applied to prediction and classification problems in many real world situations. However, a drawback of this type of neural network is that it requires a full set of input data, and real world data is seldom complete. We have investigated two ways of dealing with incomplete data — network reduction using multiple neural network classifiers, and value substitution using estimated values from predictor networks — and compared their performance with an induction method. On a thyroid disease database collected in a clinical situation, we found that the network reduction method was superior. We conclude that network reduction can be a useful method for dealing with missing values in diagnostic systems based on backpropagation neural networks.
KeywordsBackpropagation Classification Decision support Neural networks Thyroid disease
Unable to display preview. Download preview PDF.
- 1.Solberg HE. Discriminant analysis. Crit Rev Clin Lab Sci 1978; 9: 209–242Google Scholar
- 2.Sharpe PK, Solberg HE, Rootwelt K, Yearworth M. Artificial neural networks in thyroid function diagnosis from in vitro laboratory tests. Clin Chem 1993; 39(11): 2248–2253Google Scholar
- 3.Schioler T, Grimson W, Sharpe Pet al. Automatic decision support based on voting by independent decision support systems. Proc 9th Int Conf Computing in Clinical Laboratories, Dublin, Ireland, 1992Google Scholar
- 4.Boswell RA. HyperNewID and NewID. Technical Report TI/P2154/RAB/4/9.2, Turing Institute, 1992Google Scholar
- 5.Grimson W. Private communicationGoogle Scholar
- 6.Vamplew P, Adams A. Missing values in a backpropagation neural net. Proc 3rd Australian Conf Neural Networks (ACNN92), Canberra, Australia, 1992; 64–66Google Scholar
- 7.Schioler T, Nolan J, McNair P. Transferability of knowledge based systems. In: Adlassnig K-Pet al (eds), Lecture Notes in Medical Informatics No. 45, Medical Informatics Europe 1991. Springer-Verlag, 394–398Google Scholar
- 8.Forsstrom J. Inductive learning of thyroid functional states using the ID3 algorithm: The effect of poor examples on the learning result. Int J Biomed Comput, 1992; 30: 57–67Google Scholar