The present work deals with the multivariate data analysis to elucidate the hypertension risk factors in children, which may contribute in
some extent to their prospective medical treatment. Results of laboratory tests together with the data obtained from medical documentation
were used to indicate as well as to predict the hypertension diagnosis in children. The best diagnostic classification outputs were obtained
using artificial neural networks, the K-th nearest neighbour technique, general and linear discriminant analysis. In contrast to the
assessment of single laboratory test results, a combination of several tests enables more comprehensive information that can help the
physician in diagnosis. This work exemplifies a possible approach to computer-aided medical diagnosis.
Key words: hypertension, prediction, classification, multivariate data analysis
|
