Application Of Genetic Algorithm BP Neural Network In The Diagnosis Of Liver Cirrhosis Clinical Stage

Objective:BP neural network based on genetic algorithm was introduced into the clinical data of cirrhosis of liver,finishing prediction analysis on cirrhosis in the diagnosis of clinical stage.The result of cirrhosis stage prediction is greatly improved by this BP neural network.Methods:Collecting nearly 10 years hospitalized cirrhosis patients data in First Affiliated Hospital of Shanxi Medical University ‘s Department of Gastroenterology from January2006 to December 2015.According to the characteristics of data,the staging data of cirrhosis were respectively analyzed by Logistic regression,BP neural network and genetic algorithm BP neural network model.The three models were compared and the appropriate model was selected to predict the cirrhosis staging data.Results:1、Modeling and prediction of Logistic regression,BP neural network and GA-BP are respectively carried out and the results show that the median of ACC reached 90% in GA-BP,while the ACC can only reach 83.33% in BP neural network model,and more higher than 46.67% in Logistic regression;the median of TPR reached 90% in GA-BP,while the TPR can only reach 83.33% in BP neural network model,and more higher than46.67% in Logistic regression;the median of TNR reached 90% in GA-BP,while the TNR can only reach 83.33% in BP neural network model,and more higher than 46.67% inLogistic regression;the median of PV+reached 95.35% in GA-BP,while the PV+can only reach 91.30% in BP neural network model,and more higher than 80% in Logistic regression;the median of PV-reached 77.80% in GA-BP,while the PV-can only reach57.10% in BP neural network model,and more higher than 19.40% in Logistic regression;the median of AUC reached 90% in GA-BP,while the AUC can only reach 83.33% in BP neural network model,and more higher than 46.67% in Logistic regression.2 、 Comparing of BP neural network and BP neural network based on genetic algorithm,the results show that the prediction accuracy is 73.33% before optimization,and the prediction accuracy can reach 90% after optimizationand.The prediction performance is improved.The other five prediction indexes are also higher after optimization.Conclusions:BP neural network was more suitable for this study data than Logistic regression.Then the genetic algorithm is used to optimize the BP neural network,which can improve the result of stage prediction and shorten the running time.It can be concluded that the prediction result of cirrhosis staging of genetic algorithm is better than BP neural network,and it’s more feasibility in the diagnosis of cirrhosis clinical stage.