Evaluation of an epidemiological data set as an example of the application of neural networks to the analysis of large medical data sets

T. Waschulzik, K. Quandt, M. Lewis, A. Hormann, R. Engelbrecht, W. Brauer

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

3 Scopus citations

Abstract

The objective of this paper is to assess the suitability of neural networks for the analysis of large medical data sets. For this purpose, we used an epidemiological data set from the Augsburg MONICA study (MONItoring trends and determinants in CArdiovascular diseases). The method of perception analysis and the interpretation of the model established by the network are explained. The results of a prediction of survival or death of patients who suffered acute myocardial infarction show that even simple perceptrons yield good classification quality. Further, the analysis of the connections within the neural network leads to information about relations between input and output data which are of interest for future epidemiological research. The results of these investigations can also be applied to other medical data sets like data on the segmentation of CTs and ECG data for the diagnosis of myocardial infarction.

Original languageEnglish
Title of host publicationArtificial Intelligence in Medicine
EditorsS. Andreassen, R. Engelbrecht, J. Wyatt
PublisherPubl by Elsevier Science Publishers B.V.
Pages466-476
Number of pages11
Volume10
ISBN (Print)905199141X
StatePublished - 1993
Externally publishedYes
EventProceedings of the 4th Conference on Artificial Intelligence in Medicine Europe - Munich, Ger
Duration: 3 Oct 19936 Oct 1993

Conference

ConferenceProceedings of the 4th Conference on Artificial Intelligence in Medicine Europe
CityMunich, Ger
Period3/10/936/10/93

Fingerprint

Dive into the research topics of 'Evaluation of an epidemiological data set as an example of the application of neural networks to the analysis of large medical data sets'. Together they form a unique fingerprint.

Cite this