Artificial neural network-based study can predict gastric cancer staging

Background/Aims: Primary gastric cancer is a multi-factorial disease comprising many low-penetrance clinicopathological factors and genetic predisposition. Preoperative prediction of tumor staging can be made by artificial neural network (ANN)-based study using clinic-pathological datasets and genetic susceptibility testing. Methodology: A hospital-based, retrospective, randomized control study was conducted for 121 patients who had recently developed primary gastric cancer. Clinical data and pathological findings were collected and genetic polymorphisms of candidate genes were evaluated. ANN-based study was conducted to predict tumor staging and to evaluate the relative impact of each factor. Results: The best training method was the quick method, which had an accuracy of 81.82%. The most important factors associated with tumor staging were age and polymorphisms of genes p21, IL-1, IL-4 and p53. Conclusions: Analysis of genetic polymorphisms of candidate genes by ANN using clinicopathological datasets is a promising method for predicting human gastric cancer staging. This strategy can identify the important genetic, clinical and pathological factors, determine their relative impact, and aid in the development of a prognostic staging system that is useful in individualized patient care.