Study On The Influencing Factors Of Regnancy Induced Hypertension

In recent years,with the frequent occurrence of pregnancy-induced hypertension,people are paying more and more attention to how to effectively prevent the occurrence of the disease,so as to avoid the threat of disease.Although pregnancy-induced hypertension is a common disease that occurs during pregnancy,due to its many triggering factors,the medical community has not been clear about its pathogenesis.Therefore,it is urgent to explore the work that affects the development of pregnancy-induced hypertension.The generalized linear model is a kind of model developed in the 1970 s.Its statistical inference theory has been basically perfected and has been widely used in many fields.Since then,relevant theories have been greatly developed.Among them,the hierarchical generalized linear model is one of the important models.The basic meaning of the hierarchical generalized linear model is that based on the generalized linear model,a prior distribution is introduced for some parameters to form different levels of the model.For example,the prior model is a truncated poisson distribution,and the typical model with a posterior distribution of binomial distribution is the binomial-Poisson model.The machine learning methods for processing classified data include support vector machine and adaptive enhancement.The main idea of support vector machine is to further optimize the interval based on the correct classification of the training data set,and when the data is linearly inseparable or non-linear,it is necessary to introduce penalty parameters and kernel functions for processing.The learning algorithm is to optimize the objective function by learning convex quadratic programming by using dual properties.Adaptive enhancement is an integrated learning method that can be used to classification problems.Its learning process is to consecutive optimize and optimize the same weak classifier,which is different from random forest integration of multiple classifiers or multiple same classifier processes.Specifically,by changing the weight of the training sample,the basic classifier is learned,and the weight of the training sample is re-given to the classifier based on the generated error,and the classifier is consecutive iterated,and finally the classifiers are linearly combined to form a final model.The study firstly screens the variables by V correlation coefficient,"entropy" correlation coefficient and Information Value(IV),and then uses the selected variables to establish a logistic regression model of pregnancy hypertension,binomial-Poisson model,support vector machine and adaboost.The model is evaluated,which based on the accuracy index of the prediction on the test data set.Then we select the model with the highest accuracy and give conclusions and recommendations based on the significant influencing factors.It is of great significance to improve the present situation of pregnancy induced hypertension.