Research And Application Of Gastric Cancer Diagnosis Prediction Model Based On Improved Deep Belief Network

As the digitization and networking develop and the artificial intelligence technology applies universally in recent years,the medical field has accumulated great amounts of data,accordingly a huge medical database has been established,which provides a good platform for the in-depth cross integration of medical diagnosis and informatics.Gastric cancer is a kind of the commonest alimentary system diseases.At present,Surgery is the only way which can cure stomach cancer.Patients with early-diagnosed can get a better prognosis through radical surgery.However,those who diagnosed with advanced gastric cancer should consider the necessity of surgery according to the cancer stage,and then cooperate with chemotherapy or radiotherapy to prolong survival time.In the prognosis of gastric cancer,clinicians often make judgments based on the patient’s tumor diameter,tumor location,depth of invasion and other factors,which have higher requirements for the doctor’s clinical experience.Scientifically and effectively assessing the survival time of patients with gastric cancer will help improve the level of medical diagnosis and therapy,and allow patients to receive treatment as soon as possible.To a certain extent,it can increase the possibility of patients being cured and reduce the waste of medical resources.Based on the characteristics of the prognostic indicators of gastric cancer patients,this thesis constructs a gastric cancer diagnosis prediction model based on an improved deep belief network,designs and implements a gastric cancer diagnosis prediction system,and provides doctors with more useful information to cut down the probability of misdiagnosis and miss-diagnosis.The detailed work is as follows:First of all,in view of the outstanding characteristics of deep belief network(DBN)for processing high-dimensional data,the thesis uses DBN to predict and analyze gastric cancer data.In order to improve the classification ability of DBN,the last layer output of DBN is taken as the input of support vector machine(SVM),and an improved deep belief network model is constructed to realize the prediction and classification of the survival rate of gastric cancer disease.Then,download the gastric cancer data from the SEER database.For the sake of enhancing the efficiency of model data mining,the thesis pre-treated the original data to obtain a standard data set that can be used in the model.The thesis uses grid search and cross-validation to optimize the hyper-parameters of the DBN-SVM model to obtain the best classification effect.Then the model used in the thesis is compared with the BP(Back Propagation)neural network and decision tree algorithm.The experimental results prove that the classification performance of the DBN-SVM model is better.Finally,in response to the actual needs of clinical medicine,on account of the gastric cancer diagnosis prediction model constructed in the thesis,the gastric cancer diagnosis prediction system was designed and developed using Java language,which provides effective information for doctors in clinical diagnosis,and also brings patients more precise treatment.