不同中医证型复发脑梗死的眼底血管特征分析及预测模型的建立

Characteristics of Retinal Vessels and Establishment of Prediction Model of Recurrent Cerebral Infarction with Different TCM Syndromes

目的:分析影响复发脑梗死的眼底血管特征参数,建立不同中医证型复发脑梗死的预测模型。方法:回顾性分析196例在本科住院的脑梗死患者的临床资料,其中初发脑梗死137例,复发脑梗死59例。采集所有患者视网膜图像后用自动化的视网膜图像分析系统进行眼底血管特征分析,建立不同中医证型脑梗死复发风险预警模型,评估该模型的预测价值,分析复发脑梗死的眼底血管影响因素。结果:利用logistic回归模型建模,对测试集样本预测,气虚血瘀型和风痰阻络型复发脑梗死模型预测正确率分别为86.3%和87.0%。结论:利用logistic回归分析建立预测模型可以预测不同中医证型脑梗死复发的风险;多种眼底血管特征与脑梗死复发相关。

Objective ：To analyze the characteristics of retinal vessels influencing recurrent cerebral infarction and establish prediction model of recurrent cerebral infarction with different TCM syndromes. Methods ： Clinical data of 196 patients with cerebral infarction were retrospectively analyzed,including 137 patients with new onset cerebral infarction and 59 patients with recurrent cerebral infarction. Retinal images of all patients were collected by automated retinal image analysis system to analyze the characteristics of retinal vessels. Prediction model of recurrent cerebral infarction with different TCM syndromes was established and the prediction value of the model was evaluated. The factors of retinal vessels influencing recmTent cerebral infarction were analyzed. Results：Using the logistic modeling regression model to predict the test samples, the prediction accuracy of recurrent cerebral infarction model with the syndrome of blood stasis due to qi deficiency and the syndrome of obstruction of collateral due to wind -phlegm were 86.3 % and 87%. Conclusion：The model established by logistic regression analysis can predict the risk of recurrent cerebral infarction with different TCM syndromes. Various characteristics of retinal vessels are associated with recurrent cerebral infarction.