Risk Predicion Model Of Ischemic Stroke Based On Data Mining And Its Application

Stroke has the characteristics of "four highs and one more" and the younger trend of the affected population,which is irreversible and seriously threatens human life and health.Among them,ischemic stroke accounts for 60%-70% of the total number of strokes.Therefore,it is important to screen individuals at high risk for ischemic stroke in advance.Based on data mining classification algorithms and their related theories,this paper constructs a risk prediction model for ischemic stroke,screens high-risk populations,and conducts health management.First,samples are obtained from partner hospitals,the collected sample data is pre-processed using related data processing techniques,and the logistic regression algorithm is used to streamline the index set for predicting ischemic stroke risk.Secondly,in order to improve the operability and practicability of the model,different ischemic stroke risk prediction models are established on the streamlined indicator set from the simple personal data set and the complex clinical data set to screen different levels Stroke at high risk.Third,in order to circumvent the limitations of a single classifier,this article comprehensively compares the advantages and disadvantages of multiple classifiers(decision trees and support vector machines),and at the same time uses multiple model evaluation methods to build a predictive model of ischemic stroke risk The performance was evaluated.Finally,in order to achieve the high-precision and high-accuracy requirements for screening high-risk stroke patients in the medical field,this paper uses a grid search algorithm to optimize the support vector machine algorithm and the random forest algorithm integrated with the decision tree,respectively.Compare the classic algorithms,and finally seek to build the best risk prediction model in all aspects.At the same time,the prediction model is applied to the health management framework of ischemic stroke.The evaluation results of the ischemic stroke risk prediction model constructed in this paper are as follows:(1)In a single classifier,the support vector machine model has the best performance.(2)In the integrated optimization model,the performance of the RF model optimized by grid search is superior to the model constructed by any singleclassifier in all aspects,and it is also superior to the constructed grid search optimized SVM model,becoming an ischemic brain Stroke risk prediction models have the best stability,the fastest learning speed,and the best classification effects.