基于多模态MRI参数构建TIA患者认知功能障碍的列线图预测模型OA

Objective:To analyze the risk factors of cognitive impairment(CD)in patients with transient ischemic attack(TIA),and to construct a nomogram prediction model of CD in patients with TIA based on multimodal magnetic resonance(MRI)parameters.Methods:A total of 600 patients with TIA who visited our hospital from December 2020 to December 2024 were selected and divided into a modeling set of 330 cases and a validation set of 270 cases.They were divided into the CD group(n=252)and the non-CD group(n=348)based on whether CD occurred.The multimodal MRI parameters of the two groups were compared.Logistic regression was used to analyze the risk factors of CD in patients with TIA.A nomogram prediction model was constructed through R software.Internal validation was conducted using the Bootstrap method,and external validation was carried out using the validation set.The goodness-of-fit test was performed using Hosmer-lemeshow,and the predictive efficacy of the Logistic regression model for CD in patients with TIA was analyzed by the receiver operating curve(ROC).Results:(1)600 patients with TIA were divided into a modeling set of 330 cases and a validation set of 270 cases.CD occurred in 252 cases(42.00％).Among them,there were 116 cases(42.96％)in the validation set and 136 cases(41.21％)in the modeling set.There was no statistically significant difference in the incidence of CD(P＞0.05)(x2=0.187,P=0.666).(2)The ADC values and WML scores of frontotemporal and basal ganglia in CD group were higher than those in non-CD group(P＜0.05)(t=15.310,16.324,31.478,P＜0.05).(3)The ROC curve showed that the ADC value of the frontial-temporal lobe[OR(95％CI):1.684(1.227～4.505)],the ADC value of the basal ganglia[OR(95％CI):1.736(1.178～4.665)],and the WML score[OR(95％CI):1.412(1.146～3.423)],age[OR(95％CI):4.202(1.767～12.591)],resting lacunar infarction[OR(95％CI):1.923(1.336～5.892)],ABCD2 grade[OR(95％CI):2.018(1.116～5.986)]was a risk factor for CD in patients with TIA(P＜0.05).(4)Hosmer-lemeshow goodness of fit test,x2=3.036 P=0.419,indicating good goodness of fit.Internal validation showed that the nomogram model diagram and calibration curve based on CD in TIA patients predicted a good consistency between the probability of CD occurrence in TIA patients and the actual incidence of CD in TIA patients.Internal validation shows that the model has good discrimination and calibration in predicting the risk of CD onset in patients with TIA.The predicted risk is in good consistency with the actual situation,and the Brier score is 0.141.The C-index was 0.915.The nomogram model graph had a good discriminatory ability in predicting the onset of CD in patients with TIA,with an AUC of 0.915(95％CI:0.829～0.943).The predictive model had high discriminative ability and predictive value.Conclusion:Increased ADC values in the frontotemporal lobe/basal ganglia and elevated WML scores are closely related to CD in TIA patients.The established nomogram model presents superior predictive performance compared with the CAIDE scale(AUC=0.82),which can provide a basis for early clinical identification of high-risk patients.