Abstract: Objective　This study aims to establish the Bayes Model of Kawasaki diseases with coronary artery lesions （CAL） in combination by Bayes method，investigate the Cognitive Model of Kawasaki diseases with CAL.Research was performed according to the clinical biomarkers of Kawasaki diseases.Methods　We obtained clinical records of 784 patients with Kawasaki disease and hospitalized in Children’s Hospital Affiliated to Chongqing Medical University from September 2014 to September 2015.These records were retrospectively analyzed,univariate analysis on influencing factor for Kawasaki disease with CAL was carried out，and the statistical significance factors were included in the Bayes Model.Results　The univariate analysis results indicated that the gender,age,duration of the fever,PLT,Hb,CRP,and Alb were risk factors for Kawasaki disease with CAL （P<0.05）.Bayes Model showed sensitivity was 58.1%,specificity was 74.4%,and coincidence rate was 71.6%.Conclusion　The major risk factors of Kawasaki disease with CAL were gender,age,duration of the fever,PLT,Hb,CRP,and Alb.The Bayes Model improved the sensitivity for predicting the Kawasaki disease children with CAL.

Key words: Kawasaki disease, Cardiovascular diseases, Bayes theorem, Cognitive map

摘要： 目的　根据川崎病临床生物指标，结合贝叶斯方法，建立川崎病并发冠状动脉损伤（CAL）的贝叶斯模型，对川崎病并发CAL的认知模型进行初步探讨。方法　回顾性分析2014年9月—2015年9月重庆医科大学附属儿童医院784例住院川崎病患儿的临床资料，对影响川崎病患儿并发CAL的因素进行单因素分析，将有统计学意义的指标纳入贝叶斯模型。结果　784例患儿中356例（45.4%）并发CAL，单因素分析显示，性别、发病年龄、热程、血小板计数（PLT）、血红蛋白（Hb）、C反应蛋白（CRP）、清蛋白（Alb）是川崎病患儿并发CAL的影响因素（P<0.05）。贝叶斯模型的灵敏度为58.1%，特异度为74.4%，符合率为71.6%。结论　川崎病并发CAL的主要危险因素有性别、发病年龄、热程、PLT、Hb、CRP、Alb,利用贝叶斯模型提高了预测川崎病患儿并发CAL的灵敏度。

关键词: 川崎病, 心血管疾病, 贝叶斯定理, 认知图