| Objective To analyze the risk factors of acute ischemic stroke in elderly T2DM patients,in order to establish and verify the risk prediction model of acute ischemic stroke in elderly T2DM patients.Methods The clinical data of patients with T2DM who were hospitalized in the Department of Geriatrics and Neurology of the Second Hospital of Lanzhou Universi ty from January 2018 to August 2021 and aged≥60 years were retrospectively analyz ed.The data set was randomly divided into a modeling cohort（75%,n=334）and a val idation cohort（25%,n=111）.Glmnet package of R software was used to screen the hi gh risk predictors of acute ischemic stroke in elderly T2DM patients by Lasso regressi on.As the penalty coefficientλgradually increases,the coefficient of independent var iables included in the model will be gradually compressed to 0.In order to achieve go od model performance and minimize the influence factors,After 80 internal cross vali dation,the optimal penalty coefficientλ（0.072780）was obtained as the optimal value,and 6 variables were finally screened out.The above variables were refitted to the Lo gistic regression model,and based on this model,the nomogram was developed throu gh the rms package of R software.Area under ROC Curve（AUC）,namely C-index,w as used to evaluate the prediction ability of the nomogram.The modeling cohort samp les were repeatedly sampled for 100 times by Bootstrap method to conduct internal ve rification of the model.the C-index was calculated to evaluate the differentiation of th e nomogram prediction model.External verification was carried out through validatio n cohort,The calibration curves of predicted results and actual results were drawn,C-i ndex,Brier score,calibration intercept and calibration slope were calculated for comp rehensive evaluation model differentiation and calibration degree.Results（1）Comparison of clinical data in a modeling cohort:Cases group（Elderly T2DM patients with acute ischemic stroke group）male,age,duration of hypertension,diabetes,hypertension,systolic blood pressure,pulse duration pressure difference,glycosylated hemoglobin,monocytes,MHR,uric acid,CIMT,carotid artery plaque area,the number of unstable plaque,internal carotid artery stenosis degree,the degree of intracranial artery stenosis were higher than the control group（simple elderly T2DM group）,the differences were statistically significant（P<0.05）,the drinking history,high-density lipoprotein and FT3 in case group were lower than those in control group,the differences were statistically significant（P<0.05）.（2）Comparison of clinical data between the modeling cohort and the validation cohort:Most of the 334 patients in the modeling cohort and 111 patients in the validation cohort had Cohen’s d values less than 0.20 and very few less than 0.50,indicating that all baseline variables were comparable between the two groups.（3）Correlation between MHR and ischemic stroke:MHR was positively correlate d with neutrophil,lymphocyte,uric acid,triglyceride,carotid plaque area,number of unstable plaque and degree of cerebral artery stenosis（P<0.05）.MHR,PLR and NL R were compared in the area under ROC curve at the same time,and it was found that MHR had the largest area under ROC curve（AUC=0.6560,95%CI:0.5977～0.7143,P<0.001）,indicating that MHR had higher diagnostic value than PLR and NLR.The op timal cut-off value of ROC curve was 1.2568,and MHR was 0.3585×10~9/mmol,with a sensitivity of 64%and specificity of 61%.（4）Screening predictors:Diabetes mellitus with hypertension course（OR=1.05,95%CI:1.0～11.10,P<0.05）,systolic blood pressure（OR=2.40,95%CI:1.74～3.40,P<0.05）,monocyte/high-density lipoprotein（OR=1.34,95%CI:1.14～1.58,P<0.05）,FT3（OR=0.36,95%CI:0.20～0.64,P<0.05）,carotid artery plaque area（OR=1.45,95%CI:1.19～1.79,P<0.05）and cerebral artery stenosis degree（OR=4.77,95%CI:2.60～9.81,P<0.05）were screened by Lasso regression.They were independent predictors of the higher ris k acute ischemic stroke.（5）Internal validation:The C-index of the modeling cohort was 0.862（95%CI:0.824～0.900）.After 100 times of bootstrap internal verification,the corrected C-index was 0.852,the intercept value was 0.48,and the sensitivity and specificity were80.1%and 76.1%.（6）External validation:The C-index of the validation cohort was 0.867（95%CI:0.803～0.932）,the intercept value was 0.41,and the sensitivity and specificity were 75.4%and 82.0%.In the external verification,the brier-Score of calibration curve was 0.149,the calibration intercept was-0.149,and the calibration slope was 1.096.（7）Analysis of other risk factors in patients with T2DM complicated with ischemic stroke:The history of hypertension,low high-density lipoproteinemia and hyperuricemia in the case group were significantly higher than those in the control group,with statistical significance（P<0.05）.Conclusion（1）Six high-risk variables affecting the occurrence of acute ischemic stroke in elderly T2DM patients were screened out by Lasso regression,including the course of diabetes mellitus combined with hypertension,systolic blood pressure,MHR,FT3,carotid artery plaque area,the brain artery stenosis degree,according to the above variables to build nomogram prediction model in elderly T2DM with acute ischemic cerebral,The internal and external validation models have good discrimination and calibration,and the accuracy is fair.（2）High level of MHR is a risk factor for acute ischemic stroke in T2DM,When MHR>0.3585×10~9/mmol,there was a better predictive value for acute ischemic stroke in elderly patients with T2DM. |
PDF Full Text Request