Guided Diagnosis And Prediction Of Breast Cancer Based On Machine Learning

Health is the basis for the all-round development of mankind.The most common cancer in the world has changed from lung cancer to breast cancer.In recent years,the increasing trend of breast cancer incidence in our country is higher than developed countries,which seriously affects women’s health.At present,one of the treatment princples for breast cancer is stil to detect as early as possible anf inhibit the development of cancer cells as early as possible.However,the pathological characteristics of breast cells are complex and diverse,which undoubtedly adds difficulty to the diagnosis of doctors.In response to the above problems,this article is based on the machine learning method,starting from the patient registration before admission,analyzing the symptoms of the patient,and correctly guiding the patient to the breast surgery for diagnosis,and then analyze and summarize the patient’s breast pathological data.So as to assist doctors make the diagnosis of breast cancer and predict the recurrence of cancer cells in confirmed patients.The research content includes:Firstly,the guided diagnosis model based on feature expansion(L-NB).Due to the wide variety of breast diseases but the sparse symptom features,it is difficult for patients to accurately determine whether their symptoms need to go to breast surgery.Therefore,this paper proposes a guided diagnosis model based on the LDA(Latent Dirichlet Allocation)topic model for feature expansion.The research of the guided model helps patients with limited medical knowledge to accurately predict whether they need to go to breast surgery for diagnosis by entering brief symptom information.The research process includes adding a custom symptom dictionary to the existing word segmentation tools for word segmentation,using the bag-of-words model to vectorize the segmented text,and adding the expanded features to the training set to train together to optimize the model accuracy.Secondly,the diagnosis and prediction model of breast cancer based on SAv NN.Due to the complex and diverse breast pathology data,this paper proposes an adaptive neural network model.The model uses the self-attention method to extract features,which solves the problem that the recurrent neural network cannot be parallelized.At the same time,it also solves the problem that the convolutional neural network can only convolve local features each time and cannot consider global features.The model uses the grid search method to adjust the relevant parameters of the neural network according to different types of data sets,so that the prediction effect of the model is optimal.Thirdly,the diagnosis and prediction model of breast cancer based on D-Adaboost.Since the traditional deep learning model is a "black box model",it requires a large amount of data to train the model to stabilize the neural network.However,the pathological data of breast diseases is complex and changeable,and the amount of data of various pathological data is not the same.In order to further improve the stability and prediction accuracy of the model,this paper combines the decision point analysis method and the integrated learning method to construct the diagnosis and prediction model of breast cancer based on D-Adaboost.The model first uses stepwise regression to extract features from the data again to obtain the best feature combination.After training different types of breast pathology data,the model obtained the best decision point and the best accuracy rate of the model corresponding to the decision point.Through the research in this article,it is found that the guided diagnosis model based on feature expansion can accurately classify patients to breast surgery based on the patient’s limited symptom information;the prediction model based on SAv NN has a higher accuracy,recall and F1 value in breast cancer diagnosis tasks;the prediction model based on D-Adaboost has better accuracy in the task of predicting breast cancer recurrence.