Based On The Prognosis Of Epithelial Ovarian Cancer Models In Artificial Intelligence Research

Objective:To analyze the related factors with prognosis in patients with serous ovarian adenocarcinoma and to set up artificial neural networks(ANN) prognosis models for predicting ovarian adenocarcinoma,in order to provide guidence for clinical therapy.Methods:The data of 92 patients with epithelial ovarian carcinoma were randomly divided into training and testing groups. The training subsets were analyzed retrospectively and Kaplan-merier survival curves, long-rank test and Cox multiple factor survival analysis were conducted with SPSS 17.0,to screen out significant single parameters and to build the ANN's model.The testing subsets were used to estimate the performance of ANN's model and Feng grading model.Results:Univaliate analysis showed that 8 of 17 indicators may affect the ovarian cancer. Selecte by the cox multivariate analysis, FIGO stage,hiatological grade, tumor residues and sensitivity to platinum in the chemotherapy were the major prognostic factors of epithelial ovarian cancer. We set up artificial neural networks(ANN) prognosis models for predicting ovarian adenocarcinoma according to the same input variables selected by multivariate analysis had a significantly higher progositic rate than Feng grading model.Conclusion:Artificial neural network used in prediction of the prognosis for ovarian adenocarcinoma have better judgment capability.There is a need for investigating it.