摘要
目的分析影响脑梗死复发的眼底血管特征,建立脑梗死复发的预测模型。方法回顾性分析104例脑梗死患者的临床资料,其中初次脑梗死79例,脑梗死复发25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,建立脑梗死复发风险预警模型,评估该模型的预测价值,分析脑梗死复发的眼底血管影响因素。结果建立脑梗死复发logistic回归预测模型:logit P=467.210×软性渗出物+117.754×脂代谢异常+2.895×(小静脉分支角度/动静脉比值)-0.986(糖尿病/小静脉分支角度)+35.802×(糖尿病/静脉分支系数)+42.314×(糖尿病/软性渗出物)+0.435×(糖尿病/体重指数)-38.197×(脂代谢异常/软性渗出物)-10.992×(高同型半胱氨酸血症/血管弯曲度),预测准确率为90.38%。软性渗出物、脂代谢异常是脑梗死复发的独立影响因素(P<0.05)。小静脉分支角度及动静脉比值、糖尿病及小静脉分支角度、糖尿病及静脉分支系数、糖尿病及软性渗出物、糖尿病及体重指数、脂代谢异常及软性渗出物、高同型半胱氨酸血症及血管弯曲度与脑梗死复发相关(P<0.05)。结论利用logistic回归分析建立预测模型可以预测脑梗死复发的风险,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 to establish the prediction model of recurrent cerebral infarction. Methods Clinical data of 104 patients with cerebral infarction were retrospectively analyzed,including 79 patients with new onset cerebral infarction and 25 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. The prediction model of recurrent cerebral infarction was established,and the prediction value of the model was evaluated. The factors of retinal vessels influencing recurrent cerebral infarction were analyzed. Results The logistic regression model for predicting recurrent cerebral infarction： logit P = 467. 210 × Exudates ＋ 117. 754 × Abnormal lipid metabolism ＋ 2. 895 ×（ venular angle / arteriolar-to-venular diameter ratio）- 0. 986（ diabetes/venular angle） ＋35. 802 ×（ diabetes/branch coefficient of vein） ＋ 42. 314 ×（ diabetes/exudates） ＋ 0. 435 ×（ diabetes/body mass index）- 38. 197 ×（ abnormal lipid metabolism/ exudates）- 10. 992 ×（ high homocysteine / vascular tortuosity）. The accuracy of the model was90. 38%. Exudates and abnormal lipid metabolism were the independent influencing factors of recurrent cerebral infarction（ P〈0. 05）. Venular angle / arteriolar-to-venular diameter ratio,diabetes / venular angle,diabetes / branch coefficient of vein,diabetes / exudates,diabetes / body mass index,abnormal lipid metabolism / exudates,high homocysteine / vascular tortuosity correlated with recurrent cerebral infarction（ P〈0. 05）.Conclusion The model established by logistic regression analysis can predict the risk of recurrent cerebral infarction. Various characteristics of retinal vessels are associated with recurrent cerebral infarction.
出处
《广西医学》
CAS
2016年第5期635-638,共4页
Guangxi Medical Journal
基金
广东省深圳市科技计划项目(JCYJ20140408152909288)
关键词
脑梗死
复发
眼底血管
风险预测
预测模型
Cerebral infarction
Recurrence
Retinal vessels
Risk prediction
Prediction model
