为了准确识别颈动脉斑块的重要生物标志物,在改进生物标志物所包含信息量的度量方法的基础上,通过向前逐步回归建立了基于变换AUC(Transformed area under curve)的颈动脉斑块稳定性预测模型.首先,在ROC(Receiver operating characteristic)空间提出变换AUC,并给出该指标在双正态分布模型和自由分布模型下的估计方法;然后,使用R统计软件,对比分析变换AUC与AUC等常用评估指标对非传统生物标志物的评估性能;最后,基于浙江医院提供的影像数据,利用变换AUC度量生物标志物的信息量,使用向前逐步回归筛选模型的方法建立高精度的颈动脉斑块稳定性预测模型.研究结果表明,该颈动脉斑块稳定性预测模型的AUC值达到0.9以上,能够准确识别斑块的稳定性,为临床医师对患者进行个性化诊疗提供更精准的参考依据.
A prediction model of carotid plaque stability based on transformed AUC
To precisely identify critical biomarkers of carotid plaques,a model for predicting carotid plaque stability based on the transformed area under curve(transformed AUC)using forward regression was built on the basis that the method for quantifying the information content within biomarkers was improved.Firstly,transformed AUC was introduced in the receiver operating characteristic(ROC)space,and the estimation methods were provided under the binormal distribution model and free distribution model,respectively.Then,R statistical software was used to compare and analyze the evaluation performance of transformed AUC index with common evaluation indices such as AUC for non-traditional biomarkers.Finally,a carotid plaque stability prediction model with high-accuracy was built by using transformed AUC to measure the information of biomarkers and stepwise forward regression based on image data provided by Zhejiang Hospital.These research findings illustrate that the AUC value of the carotid plaque stability prediction model is above 0.9,indicating the model can accurately identify the plaque stability and provide more precise reference to clinicians for personalized diagnosis and treatment decisions.
NETL
NSTL
万方数据
