Schoeller Junkmann Preis 2000
Dr. Klaus Prank
Abteilung Klinische Endokrinologie, Medizinische Hochschule Hannover
Non-linear analysis of phenotype-genotype mapping: predicting heterozygous mutations in the 21-hydroxylase gene from serum steroid profiles
Non-linear relations between multiple biochemical parameters are the basis for the diagnosis of many diseases, when traditional linear analytical methods are not reliable predictors. Novel non-linear techniques are increasingly used to improve the diagnostic accuracy of automated data interpretation, such as the analysis of electrophysiological data from ECG and EEG leading to the diagnosis of acute myocardial infarction or of certain forms of epilepsy.
Our objective was to predict the genotype from complex biochemical data by comparing the performance of experienced clinicians to traditional linear analysis, and to novel non-linear analytical methods. As a model, we used a well defined set of interconnected data consisting of unstimulated serum levels of five steroid intermediates assessed in 32 subjects heterozygous for a mutation of the 21-hydroxylase gene (CYP21B) and in 16 healthy controls.
The genetic alteration was predicted from the pattern of steroid levels with an accuracy of 39% by clinicians and of 72% by linear analysis. In contrast, non-linear analysis, such as artificial neural networks and nearest neighbor classifiers, yielded a significantly higher accuracy between 80% and 90%. These results support the potential of non-linear analysis to classify complex biochemical patterns from well defined data sets and may have general implications for the standadized diagnosis from subtle changes in multivariate biochemical data especially for the less experienced.