Learning from Noisy Medical Data: A Comparative Study Based on a Real Diagnostic Problem
Clinicians routinely collect extensive case histories on their patients and in certain medical domains this data may be supplemented with confirmed or “working” diagnosis obtained by patient follow-up. The possibility of using such datasets as a source of ‘knowledge’ for diagnostic systems has been a goal of many research studies.
This paper reports on the application of five approaches: rule induction, neural networks, statistically based diagnostic trees, Bayes discriminants and logistic models; to the construction of diagnostic aids based on a noisy, mainly categorical, medical dataset giving the clinical presentation of patients with either Multiple Sclerosis or Cerebrovascular(Vascular) Disease who have been referred for Magnetic Resonance Imaging.
The procedures investigated gave very similar results in terms of overall diagnostic performance although the ‘format’ of the resulting diagnostic aids was very different. The use of Multiple Correspondence Analysis as a preparatory technique, to remove noisy variables, proved very successful in identifying a smaller subset of items that were more amenable to ‘automated’ techniques such as neural networks/rule induction and also assisted in the selection of variables for statistical discrimination.
KeywordsMultiple Sclerosis Multiple Correspondence Analysis Multiple Episode Rule Induction Internal Auditory Canal
Unable to display preview. Download preview PDF.
- 2.Ledley RS and Lusted 1.13 Reasoning foundations of medical diagnosis, Science, 130 (1959), 9–21.Google Scholar
- 3.de Dombal FT, Leaper DJ, Staniland JR, McCann AP and Horrocks JC: Human and computer-aided diagnosis of abdominal pain; tùrther report with emphasis on performance of clinicians, BMJ, I (1972), 376 - S0.Google Scholar
- 5.Spiegelhalter DJ and Knill-Jones RP Statistical and knowledge based applications to clinical decision support systems with an application in gastroenterology, JRSS, 147 (1984)Google Scholar
- 8.du Boulay GH, Field B, Teather BA, Teather D and Plummer D: The extraction of expert knowledge for MR image acquisition from the published literature, Rivista di Neuroradiologia, 5 (1992), 473–482.Google Scholar
- 9.Enzmann DR and O’Donohue J: Optimising MR imaging for detecting small tumours in the cerehellopontine angle and internal auditory canal, A.INR, 8 (1987), 99–106.Google Scholar
- 10.Pojunas KW, Danials DL, Williams AL, Naughton VM MR imaging of prolactin secreting micro-adenomas, AJNR, 7 (I 986 ), 209–213Google Scholar
- 12.Gregson M, John R, Teather BA and Thompson R Practical issues in the application of neural networks to the differential diagnosis of brain disease, Proc. IEE International Conference on Neural Networks and Expert Systems in Medicine and Healthcare, Plymouth (1994).Google Scholar
- 14.Teather D: Diagnosis–methods and analysis, Bulletin of IMA, 10 (1974), 37–41.Google Scholar
- 16.Benzécri JP et al: L’Analyse des données Tome I - La taxinomie. Tonne 2 - L’Analyse des correspondances, Dunod, ParisGoogle Scholar
- 17.Greenacre MJ: Correspondence analysis in practice, Academic Press, London 1993Google Scholar
- 18.Greenacre MJ SimCA Version 2 - Personal Computer Software for Correspondence Analysis, User Manual, 1990.Google Scholar