Artificial Neural Network Model for Prediction of Childhood Allergic Asthma Using Single Nucleotide Polymorphism Data

摘要

Screening of various gene markers (e.g., microsatellite polymorphisms, variable number of tandem repeats, single nucleotide polymorphism (SNP)) and correlation between these markers and pathogenesis of allergic disease have been studied. However, there have been no proposed prediction models with high accuracy using such factors. In order to predict development of disease, we applied a knowledge processing method based on an artificial neural network (ANN) model to diagnostic prediction of 172 subjects with childhood allergic asthma (CAA) and 172 healthy subjects, using 25 SNPs in 17 genes. SNP data was kindly provided by the ethics committees of Tohoku University and RIKEN. For comparison with ANN, we also used logistic regression (LR) analysis, which is currently used to analyze medical statistics. In order to selectively identify causal SNPs, a parameter decreasing method (PDM) was used in the ANN modeling. Associations between combinations of significant SNPs and CAA pathogenesis were investigated. We found that the ANN was able to represent associations between CAA and 2 or 3-SNP combinations using complicated nonlinear relations. Thus, the ANN can be used to characterize development of complex diseases caused by multiple factors.