A Heart Attack Predictive Model And System Within Stroke Patients

In clinical treatment,the incidence of stroke patients with heart attack is as high as 30%,the fatality rate is as high as 54%,and the early prediction of stroke and heart attack is extremely poor.If the heart attack can be predicted in advance from other daily detection indicators,so that it can be monitored and treated in advance to reduce the mortality of stroke patients with heart attack,how to predict heart attack in the data of stroke patients is a huge challenge.Because patients with heart attack account for a small proportion of stroke patients,the data on whether stroke patients have heart attack is extremely unbalanced.This paper proposes an imbalanced data processing algorithm for stroke patients.Based on this algorithm,a prediction model of stroke patients with heart attack is constructed,and a system tool for stroke patients’ heart attack prediction is developed based on this model.The specific research content is as follows:1.This paper combine random undersampling,clustering and oversampling and design a processing algorithm for imbalanced data which is called undersamplingclustering-oversampling algorithm(referred to as UCO algorithm).The UCO algorithm inputs imbalanced data and outputs balanced data.Using these data,machine learning models can be effectively trained to predict whether stroke patients have heart attack.2.A stroke heart attack prediction model UCO random forest model(UCO(120)_RF model for short)was designed based on UCO(120),and this model was used in MIMIC-III,the American clinical critical care medicine database.The experimental results showed that the accuracy of the model It is 70.29%and the precision is 70.05%,which can accurately predict whether stroke patients will have heart attack.3.Based on the UCO(120)_RF model,this paper designs and develops a predictive system tool for stroke patients with heart attack.This tool can predict whether stroke patients have heart attack,input the patient’s clinical test data,and automatically generate Forecast results and forecast reports.Experiments show that the system can predict the data of stroke patients in real time,helping doctors to make decisions in advance for clinical treatment.