Study On The Prognosis Model For Lung Cancer Patients Based On Machine Learning

Cancer is a serious harm to the health of human life,lung cancer is a common kind of cancer in life,but also a serious impact on human health,a cancer.The treatment of lung cancer has become a hot topic in the medical profession.There are scholars to do on lung cancer pathology,lung cancer diagnosis,lung cancer treatment and other aspects of lung cancer research.In the study,more and more scholars use machine learning algorithm to model,through the analysis of existing data to do research.Usually using decision tree,Logistic regression,neural network model and other machine learning methods to study,and the results are very valuable.SVM is often used as a classifier in industrial research and has been used in medical research.On the prognosis of lung cancer model,some scholars use artificial neural network method and the application of six indicators of joint detection to explore the survival of patients with lung cancer,the experimental results can reach 88.3% accuracy,but there are less data characteristics,the total amount of data less,Modeling methods and other shortcomings.Some scholars have used SEER dataset to apply decision tree,Logistic regression and neural network model to study the prognosis of breast cancer patients,but there are some shortcomings in data preprocessing.This paper mainly does the following:(1)The SEER dataset used in this paper has 155 feature fields.For the characteristics of the SEER dataset,define the feature selection methods that apply to SEER datasets or other large medical data sets.Finally,from the 155 field final selection of 27 fields.And through the experimental results we can see that the custom feature selection method to improve the performance of the model.(2)In this paper,the prognostic model of lung cancer patients was established by using the decision tree,neural network,SVM and Logistic regression model.The model of each model was studied by changing the input characteristics and changing the parameters of the algorithm.The optimal input of each model algorithm is obtained,and the optimal model of each model algorithm is established.The neural network model is the best modeling algorithm in four algorithms.In this paper,the input characteristics of the model can be improved by integrating the importance of the four model algorithms.V1,V24,V10,V108,V136,V8,V117,V97,V21,V27 as input features,using MLP neural network modeling,which has two hidden layers,and hidden layer 1 unit number is 2,hidden layer 2 is 1,the accuracy rate is 90.423%,which is the optimal model of this paper.(3)In this paper,the depth of neural network is used to establish the prognosis model of lung cancer patients based on depth neural network.The absolute accuracy of the model is obtained and the accuracy of the neural network model is lower than that of the neural network model.However,it is found that the accuracy of the depth neural network is better than that of the neural network model when the number of input feature groups is increased.Therefore,the depth neural network is more suitable for the input feature,and the depth neural network is dependent on the preprocessing less.