摘要
针对兔胸部创伤后血液凝血功能检测数据,从变量检验入手,建立了基于Logistic回归模型的兔血液凝血功能诊断模型.结果表明模型的诊断正确率均在90%以上.为了降低样本集的划分对诊断正确率的影响,本文利用jackknife检验法思想和随机抽样技术,进一步对模型的有效性进行了评估分析并且证实了剔除纤维蛋白原(fibrinogen,FIB)指标会显著降低诊断正确率.
According to the test data of blood coagulation function in rabbits after chest trauma,beginning with the variable examination,this paper established the Logistic regression model for blood coagulation function diagnosis in rabbits.The results showed that the correct diagnostic rate of the samples is more than 90%.In order to weaken the impact of dividing samples set on classification accuracy,the validity of the model was assessed by the idea of jackknife test and the random-sampling technique.Meanwhile,the results confirmed that the elimination of fibrinogen index will significantly reduce the rate of correct diagnosis.
出处
《生物数学学报》
2015年第2期267-272,共6页
Journal of Biomathematics
基金
国家自然科学基金项目(11271369)资助
