Analysis Of Clinical Features Of Transient Symptomatic Cerebral Infarction(TSI) And Establishment Of Risk Prediction Model

Objective:1.To analyze and explore the relevant clinical features of transient symptomatic cerebral infarction.2.To establish and evaluate a risk-predicting model for transient symptomatic cerebral infarction.Methods:A retrospective study was performed on patients with transient ischemic attack as admitting diagnosis from March 2014 to October 2019 in the Department of Neurology,Shanghai Tenth People’s Hospital.Then,we divided the patients into the TIA group(DWI negative group)and the TSI group(DWI positive group)on the basis of DWI results.We summarized the clinical features of TSI and established a TSI risk-predicting model using Logistic regression equation.Meanwhile,the model was visually presented in the form of nomogram,and the internal verification was conducted by Bootstrap method to evaluate the discrimination and calibration ability of nomogram.Results:A total of 230 patients were included in this study.Of which,135 were TIA,and95 were TSI group.The main mechanism of TSI were large artery atherosclerosis.The majority of lesions were single and located in frontal lobe and basal ganglia.We found that speech,smoking,uric acid,low density lipoprotein cholesterol,brain natriuretic peptide precursors and ABCD3 scores could be included as predictors(P<0.05).A risk-predicting model based on logistic regression equation was established with four significant variables(smoking history,low density lipoprotein cholesterol,brain natriuretic peptide precursor and ABCD3 score).The AUC of the prediction model equation was 0.762(95%CI 0.701-0.823),the specificity and sensitivity were 59.3%and 82.1%,respectively.Hosmer-lemeshow test results showed thatχ~2=5.027,P=0.755>0.05,suggesting that the predicted probability of the model was in good agreement with the actual observation probability and had a good calibration ability.We constructed the nomogram(ABCD3-SLOPE score)to visually present the prediction model equation.The Bootstrap method was used for internal verification.The C-statistic of the nomogram is 0.77(95%CI 0.70-0.83).The calibration curve shows that the risk probability predicted by the nomogram is in good consistency with the actual observation probability,and the nomogram has a good degree of discrimination and calibration.Conclusion:1.The mechanism of TSI patients was mainly large artery atherosclerosis.The lesions showed single,and lacunar lesions with the most involvement of frontal and basal ganglia.2.Independent risk variables for TSI were speech problem,smoking history,hyperuricemia,low density lipoprotein elevated,high brain natriuretic peptide precursors and high ABCD3 scores.3.The prediction model equation based on Logistic regression and the nomogram(ABCD3-SLOPE score)that could present the prediction model intuitively and concisely could both predict the occurrence risk of TSI well.