Application Of Parametric Method And Non-parametric Method In Estimation Of Area Under ROC Curve

Objective To elucidate and compare the parametric method and non-parametricmethod in estimation of the area under ROC curve, so as to provide a basis fortheir application in diagnosis assessment.Methods The areas under ROC curves were estimated by parametric method ofbinomial model and by non-parametric method of Hanley and McNeil. Bothmethods were employed in the diagnostic tests of Laryngeal Carcinoma andOvarian Carcinoma, and the result was compared.Results Both parametric method and non-parametric method demonstratedDoppler B ultra had higher diagnosis value in the diagnostic tests of LaryngealCarcinoma; Both methods demonstrated CA125, RMI1, RMI2 and RMB hadmedium diagnosis value in the diagnostic tests of Metrocarcinoma. The areasunder ROC curves estimating by two methods was similar in the diagnostic ofLaryngeal Carcinoma.,but it was different in the diagnostic tests ofMetrocarcinoma.Conclusion Both parametric method and non-parametric method can bechosen to estimate the area under the ROC curve of binomial distribution data;as to skewness distribution data,both methods can be applied to big sample sizedata; non-parametric method is better to small sample size data.