Studies On Diagnosis Of Cardiovascular Diseases Based On Data Mining

On the one hand,as an ageing population and the rapid development of social economy in our country,the heart diseased increased rapidly year by year,on the other hand,with the rapid development of medical technology,the modern medical instrument can realize all-round real-time monitoring for patients have a lot of monitoring data,the massive medical data processing is the future for the development of medical practical problems,and based on the deep study of data mining technology is one of the effective methods to solve the problem.However,how to mine knowledge from the large amount of data,and apply to medical diagnosis practice,at the same time,on the basis of guarantee the accuracy of the diagnosis,improve the data processing speed is the important research contents of need.Extraction and intelligent diagnosis based on the characteristics of the automatic classification technology two aspects,in view of the traditional feature extraction method depends on the defects of human experience,using the knowledge of unsupervised learning method to extract data automatically,and through the parameter adaptive genetic programming method to optimize feature,on this basis,the research Softmax regression model,using the advantages of simple structure to realize the rapid and effective fault classification.The detailed research contents are summarized as follows:(1)Aimed at the existing empirical knowledge based on the problem of how to improve the effect of feature extraction,feature extraction method based on genetic programming,and for key parameters in genetic programming algorithm,the initial value selection problem,put forward the key parameters of genetic programming algorithm ofadaptive adjustment,avoid the values of initial parameters on the result of optimization,and improve the efficiency of genetic programming algorithm optimization,finally through the symbolic regression experiment validates the effectiveness of the proposed method.(2)In the event of the lack of signal prior knowledge of the problems in the process of feature extraction,the proposed neural network based on the coding of unsupervised feature extraction method,using the characteristics of network hidden unit number less,through transferring data features to restore the process,realizing the data in the extraction of implicit knowledge,at the same time,combining the experimental validation of the proposed fast determine the validity of the method of network hidden layer unit number.(3)Due to the lack of rapidness,effectiveness in medical diagnosis,fault classification problem,putting forward the classification method based on Softmax regression model,the structure of the model is simple to avoid tedious training process parameters,at the same time,maximizing the sample data,increasing the probability of correct classification is learning goals which enables the model to focus more on data category information,reducing the interference of irrelevant information,improving the accuracy of the classification results.Finally,the experiment verifies the performance of the proposed classification method.(4)The above method was applied to medical diagnosis of heart disease,using the adaptive genetic programming method for patients with electrocardiogram(ECG)data feature extraction,at the same time,the use of the ECG in patients with 10000 pairs of sample data to construct an undertaking training since the encoding neural network,the situation of its simulation processing huge amounts of data,and by using the coding characteristics of neural network to extract the vibration data to extract the characteristics of the input to the Softmax regression model for training and diagnosis,in order to verify the validity of the method.