Research On Assistant Diagnosis Of Asthma Based On Machine Learning

In recent years,artificial intelligence information technology has continued to develop,and intelligent auxiliary systems have emerged.In modern medicine,doctors use their rich work experience and medical knowledge reserves to diagnose and treat patients,and artificial intelligence technology is increasingly applied to medicine to help doctors perform auxiliary diagnosis.The article mainly studies the auxiliary diagnosis method of asthma based on machine learning,which mainly includes the classification prediction of the yin positivity of palm prints,one of the main symptoms of asthma,and the classification of asthma symptoms and syndrome types.The main work content is as follows:The yin and yang features of the thenar palm prints are directly related to the prevalence of asthma.Therefore,this paper proposes the inception V3 model based on the optimized loss function to recognize the two types of thenar palm prints in the palm pictures.First use the Label Img tool to annotate the palmprint data of the big thenar,and use the data enhancement method to blur,flip,and zoom the image data to expand the data volume and further improve the performance of the network;optimize the loss function in the model,So that the final model can have a certain performance in cohesion.The experimental results show that the inception V3 model after optimizing the loss function has higher efficiency and higher recognition accuracy,and the accuracy of the negative and positive recognition of the asthma symptoms-thenar palm prints reaches90.2%;This paper introduces the data set used in this study and its related standardization process.This data set comes from the hospital information management system of Qingdao Haici hospital and Internet resource crawler data.Due to the complexity of TCM asthma medical record data,it needs to be solved one by one.Therefore,we need to preprocess the TCM medical record data of asthma first.In this step,we will use Python data analysis package and other tools,and further quantify the text information into data information that can be used for model training through map function,so as to eliminate the differences caused by different dimensions;In order to solve the problems of traditional algorithm feature extraction,low accuracy and prone to overfitting,this paper proposes an improved deep belief network(DBN)model.It introduces the use of an improved deep belief network model to train the TCM asthma medical record data mentioned in the previous chapter,so as to realize the task of syndrome classification.First introduced the improvement of the DBN model,that is,the traditional DBN model is improved by cascading SVM classifiers,the training data is input to the improved DBN model,and continuous training,iteration,optimization,and experimental results are analyzed.,Determine the best network model architecture and hyperparameter settings,and test the improved model on the test set to test the classification results of asthma syndromes,and compare the final results with other models at the evaluation index level for analysis and research.