Abstract
Recent studies have suggested that not only genetic factors such as single nucleotides polymorphisms (SNPs) but also environmental factors such as smoking, alcohol consumption, and intake of salty food affect the development of heterogeneous diseases. Revealing the polymorphisms that may contribute to Helicobacter pylori infection as a consequence of specific environmental factors will help in developing personalized medicine. Eighty-four H. pylori infection-positive and 84 H. pylori infection-negative subjects were analyzed using the data for 37 different polymorphisms with a machine learning method—artificial neural network (ANN) with the exhaustive combination search method. The constructed ANN model for H. pylori infection exhibited greater performance than the logistic regression (LR) model. ANN modeling was also separately applied to non-smokers and smokers data sets. The result implies that the different polymorphisms can be risk factors under the influence of specific environmental factors. In addition, the result also indicated that the susceptibility to H. pylori infection can be lowered by giving up smoking even though subjects with risk factor for smokers. The reported polymorphisms that are related to H. pylori infection were automatically identified using our method, without any prior knowledge. The identified polymorphisms affected the infection when present as combinations. This method will be a useful tool for the analysis of risk factors against multifactorial diseases.
