决策树模型在发热伴血小板减少综合征辅助诊断中的应用  

A decision tree for the diagnosis of severe fever with thrombocytopenia syndrome

作　　者：王涛[1] 周云平[1] 张艳[1] 张晓梅[2] 姜晓林[2] 王玲[1] 王显军[2] 丁淑军[2] 

机构地区：[1]淄博市疾病预防控制中心,山东淄博255026 [2]山东省疾病预防控制中心,山东济南250014

出　　处：《现代预防医学》2016年第16期2881-2884,2896,共5页Modern Preventive Medicine

基　　金：国家自然科学基金(81102171);山东省医药卫生科技发展计划项目(2011HZ055);山东省科技发展项目(2012GHZ30031);山东省自然科学基金(ZR2014HP030)

摘　　要：目的比较实验室确诊的发热伴血小板减少综合征病例与实验室排除的疑似病例的临床与流行病学特征,并初步探讨决策树模型在发热伴血小板减少综合征辅助诊断中的应用。方法采用描述性流行病学方法分析淄博市2011年-2014年报告的发热伴血小板减少综合征确诊及疑似病例的临床表现及流行病学特征等,两者的比较采用χ~2检验或Fisher精确检验。应用决策树模型对发热伴血小板减少综合征确诊及疑似病例进行判别分析。结果发热伴血小板减少综合征确诊和疑似病例在年龄分布、发病月份分布、有无腹泻症状、发病前2周是否割草的差异有统计学意义。决策树模型的灵敏度为0.798,特异度为0.635,ROC曲线下面积0.715。结论对于发热等临床表现且外周血血小板和白细胞降低者,决策树模型表明,如果年龄≥50岁,有腹泻症状,且有乏力症状,则判别为确诊病例。决策树模型简单、易懂,为发热伴血小板减少综合征的辅助诊断提供强有力的决策支持。Objective The aim of this study was to compare the clinical and prevalence features between laboratory confirmed and suspected severe fever with thrombocytopenia syndrome cases and explore the application of decision tree model in the diagnosis of severe fever with thrombocytopenia syndrome. Methods Descriptive epidemiology methods were used to analyze the clinical and prevalence features between laboratory confirmed and suspected severe fever with thrombocytopenia syndrome cases reported in Zibo from 2011 to 2014. Two test or Fisher＇s exact test was conducted to compare the differences.The decision tree model was applied for discrimination analysis. Results There were statistically significant differences in age,months of incidence, diarrhea and mowing. The sensitivity, specialty and area under receiver operating curve（ROC） were0.798, 0.635 and 0.715, respectively. Conclusion The decision tree model suggested that cases with fever, thrombocytopenia,leukocytopenia, ages≥50, diarrhea and fatigue were associated with laboratory confirmed cases. The decision tree model was easy to understand and can provide strong evidence for the diagnosis of severe fever with thrombocytopenia syndrome.

关 键 词：发热伴血小板减少综合征 决策树 诊断 

分 类 号：R181.2[医药卫生—流行病学] R183.5[医药卫生—公共卫生与预防医学]