Distribution Of Cerebrovascular Stenosis And Discriminant Model Research

Background:Ischemic cerebrovascular disease(ICD)is a common clinical cerebrovascular disease.Every year about 1 million people die of cerebrovascular disease.It has the characteristics of high morbidity and mortality,which brings heavy burden to families and society.Cerebrovascular stenosis is the main cause of ischemic cerebrovascular disease.Clinical diagnosis of cerebrovascular stenosis is mainly made by MRA(Magnetic Resonance Angiography)and DSA(Digital Subtraction Angiography).The specificity of MRA is very high,up to 95%,so the patients without cerebrovascular stenosis can be excluded basically.However,the sensitivity of MRA is not high,only 62%-79%.Therefore,MRA examination for people with cerebrovascular stenosis can not be confirmed.Misdiagnosis exists,and the price is very expensive.The sensitivity and specificity of DSA are very high.DSA is the gold standard for the diagnosis of cerebrovascular stenosis.Therefore,DSA examination for patients with cerebrovascular stenosis can be confirmed.However,DSA is invasive,riskier and more expensive.Therefore,MRA and DSA are not suitable for screening patients with cerebrovascular stenosis in the general population.The purpose of this study is to explore a screening technique for cerebrovascular stenosis based on economic,fast and common examination indicators such as smoking,drinking,hypertension and high density lipoprotein,and to identify patients with cerebrovascular stenosis in the general population by establishing different discriminant models for cerebrovascular stenosis.Early interventions such as lifestyle interventions,drug treatments and stent placement are necessary.It has important public health significance for preventing ischemic cerebrovascular disease such as stroke.Materials and Methods:Based on the Shandong Multi-Center Healthcare Big Data Platform,this study selected 1118 people without cerebrovascular stenosis by MRA examination from physical examiners and 1329 patients with clear cerebrovascular stenosis by DSA examination from hospitalized patients.We select 1118 people without cerebrovascular stenosis as control group and 1329 patients with clear cerebrovascular stenosis as case group to constitute research samples.Distribution of the site of cerebrovascular stenosis was described in patients.We used economic and fast examination indicators to construct the logistic discriminant model and random forest discriminant model of cerebrovascular stenosis in male and female.AUC was used to evaluate models and we compared the discriminant effect and stability of the two models.Results:1.During the 1329 cerebrovascular stenosis patients diagnosed by DSA,985(74.12%)were males and 344(74.12%)were females.Internal carotid artery,vertebral artery and middle cerebral artery were detected more.During the cerebrovascular stenosis patients diagnosed by DSA,more than half of the differences in cerebrovascular stenosis between different genders and age groups were statistically significant.There were more people with multiple cerebrovascular stenosis than those with single cerebrovascular stenosis.The proportion of severe stenosis and occlusion was higher than that of mild and moderate stenosis.2.In the logistic discriminant model of cerebrovascular stenosis,eight discriminant factors were enrolled in male including age,aspartate transaminase,albumin,high-density lipoprotein,neutrophil value,hypertension,smoking and drinking.Five discriminant factors were enrolled in female including albumin,high-density lipoprotein,fasting blood glucose and hypertension.The high-density lipoprotein,albumin,hypertension and neutrophil value were common enrolled discriminant factors in male and female.The area under the ROC curve(AUC)was 0.927 and 0.888 for men and women,respectively.3.In the random forest discriminant model of cerebrovascular stenosis,eight discriminant factors were enrolled in male including albumin,high-density lipoprotein,age,drinking,hypertension,total cholesterol,uric acid and neutrophil ratio.Nine discriminant factors were enrolled in female including high-density lipoprotein,albumin,age,triglycerides,total cholesterol,hemoglobin,platelet,neutrophil ratio and low-density lipoprotein.The age,high-density lipoprotein,albumin,neutrophil ratio and total cholesterol were common enrolled discriminant factors in male and female.The area under the ROC curve(AUC)was 0.948 and 0.810 for males and females,respectively.Conclusions:1.The high incidence sites of cerebrovascular stenosis were internal carotid artery,vertebral artery and middle cerebral artery.There were more people with multiple cerebrovascular stenosis than those with single cerebrovascular stenosis.The proportion of severe stenosis and occlusion was higher than that of mild and moderate stenosis.2.Logistic discriminant model and random forest discriminant model of cerebrovascular stenosis showed that albumin,hypertension and high-density lipoprotein were common enrolled discriminant factors and the discriminant effect and stability of the two models are ideal.3.Rapid and efficient discriminant models of cerebrovascular stenosis were established by using simple and cheap routine examination indicators,which provide a convenient tool for identification of cerebrovascular stenosis in general population.We found that in comparison of the discriminant effects of two models of cerebrovascular stenosis,the discriminant effect of random forest discriminant model of cerebrovascular stenosis in male is better than that in logistic discriminant model of cerebrovascular stenosis,while the discriminant effect of logistic discriminant model of cerebrovascular stenosis in female is better than that in random forest discriminant model of cerebrovascular stenosis.The discriminant effect of discriminant model of cerebrovascular stenosis in male is better than that in female.