Application of interactive multivariate data visualisation to the analysis of patient findings in metabolic research

摘要

In this paper, methods for multivariate data projection, the resulting feature map display, and interactive visual analysis are applied in an interdisciplinary cooperation for the analysis of patient findings in metabolic research. From a medical database collected in corresponding research, data sets were subject to preprocessing, projection, and visual analysis. Globally significant parameters could be identified by this work, that coincide with certain patient groups, e.g. characterised by the attributes of sex and diabetes. Further, the visualisation techniques opened a way to find local correlations and interdependencies between parameters within subgroups. The medical analysis is supported by a recent dedicated PC-based tool that allows the processing and interactive visualisation of the medical databases by means of novel features and techniques.