Establishment And Evaluation Of Clinical Prediction Model For The Onset Of Cerebral Small Vessel Disease

Objective:Cerebral small vessel disease(CSVD)usually starts insidiously,which can lead to abnormal mental behavior,cognitive decline,movement disorder and incontinence,etc.If the diagnosis and treatment are not timely,the quality of life of patients can be seriously affected.Based on clinical data,this study respectively collected CSVD patients as the experimental group and non-CSVD subjects as the control group and healthy people as the normal control group,and established a clinical prediction model by analyzing the relevant factors affecting the onset of CSVD.The prediction model was used to further study the risk factors of CSVD and guide the diagnosis,treatment and prognosis of the disease.Subjects and methods:A total of 170 inpatients in the Department of Neurology,in the Department of Neurology,China-Japan Union Hospital of Jilin University from May 2019 to October 2019 were collected and set as group A,while 55 inpatients from November2019 to February 2020 were collected and set as group B.According to the diagnostic criteria of CSVD,the subjects in group A were divided into 70 CSVD patients,100 normal controls,and 55 subjects in group B,among which 30 were CSVD subjects and 25 were normal controls.Collect population information data,past medical history,laboratory results and radiological examination data,will be A set of objects by Logistic regression analysis method of single factor and multiple factors regression analysis,finally to determine independent risk factors(P < 0.05)build A clinical prediction model to build CSVD incidence forecast model,and based on the degree of differentiation,calibration and clinical practicability of the model for effectiveness evaluation,in order to verify the feasibility of the mode.Results:Gender,age,history of hypertension,history of diabetes,history of smoking,history of alcohol consumption,history of atherosclerosis,triglycerides,total cholesterol,low density lipoprotein,homocysteine,uric acid,c-reactive protein and cystatin were used as independent variables.Gender(P = 0.818),age(P = 0.004),smoking habit(P = 0.452),drinking habits(P = 0.829),hypertension(P = 0.00),diabetes(P = 0.019),hardening of the arteries(P = 0.013),TG(P = 0.076),the TC(P =0.066),LDL-C(P = 0.037),Hcy(P = 0.00),UA(P = 0.026)and CRP(P = 0.075),Cys-C(P = < 0.00),Among them,the P values of age,hypertension,diabetes,atherosclerosis,ldl-c,HCY,UA and cys-c were all less than 0.05,with statistical significance.Age,hypertension,diabetes,arteriosclerosis,ldl-c,HCY,UA and cys-c were analyzed by multivariate regression.After multivariate regression,age(P=0.021),history of hypertension(P=0.022),Hcy(P=0.002),and cys-c(P=0.002)variables showed statistically significant differences in the subjects with or without cerebrovascular disease.The coefficients are extracted to form the prediction model formula.The current area of the ROC curve was AUC=0.793,and the model had good discrimination.Calibration scatter plot was used to evaluate the Calibration of the model.The data in group B was used for accurate clinical evaluation,and the sensitivity,specificity,accuracy,positive predictive value and negative predictive value were 84%,80%,90%,77% and 85% respectively.Conclusion:1.Age,history of hypertension,history of diabetes,arteriosclerosis,low density lipoprotein,homocysteine,uric acid and cystatin-C were all related to the pathogenesis of small brain vessel disease;2.Aged,history of hypertension,elevated homocysteine and cystatin-C were independent risk factors for the onset of cerebrovascular disease;3.The clinical prediction model can accurately predict the incidence rate of thesubjects,providing better clinical decision-making for the diagnosis and treatment of CSVD.