A Bayesian semiparametric approach to correlated ROC surfaces with stochastic order constraints

Reconsidering evidence-based management of endometriosis

Background: Without an adequate animal model permitting experiments the pathophysiology of endometriosis remains unclear and without a non-invasive diagnosis, information is limited to symptomatic women. Lesions are macroscopically and biochemically variable. Hormonal medical therapy cannot be blinded since recognised by the patient and the evidence of extensive surgery is limited because of the combination of low numbers of interventions of variable difficulty with variable surgical skills. Experience is spread among specialists in imaging, medical therapy, infertility, pain and surgery. In addition, the limitations of traditional statistics and p-values to interpret results and the complementarity of Bayesian inference should be realised. Objectives: To review and discuss evidence in endometriosis management Materials and Methods: A PubMed search for blinded randomised controlled trials in endometriosis. Results: Good-quality evidence is limited in endometriosis. Conclusions: Clinical experience remains undervalued especially for surgery. What is new? Evidence-based medicine should integrate traditional statistical analysis and the limitations of P-values, with the complementary Bayesian inference which is predictive and sequential and more like clinical medicine. Since clinical experience is important for grading evidence, specific experience in the different disciplines of endometriosis should be used to judge trial designs and results. Finally, clinical medicine can be considered as a series of experiments controlled by the outcome. Therefore, the clinical opinion of many has more value than an opinion.

Discriminatory Capacity of Prenatal Ultrasound Measures for Large-for-Gestational-Age Birth: A Bayesian Approach to ROC Analysis Using Placement Values

Predicting large fetuses at birth is of great interest to obstetricians. Using an NICHD Scandinavian Study that collected longitudinal ultrasound examination data during pregnancy, we estimate diagnostic accuracy parameters of estimated fetal weight (EFW) at various times during pregnancy in predicting large-for-gestational-age. We adopt a placement value based Bayesian regression model with random effects to estimate ROC curves. The use of placement values allows us to model covariate effects directly on the ROC curves and the adoption of a Bayesian approach accommodates the a priori constraint that an ROC curve of EFW near delivery should dominate another further away. The proposed methodology is shown to perform better than some alternative approaches in simulations and its application to the Scandinavian Study data suggests that diagnostic accuracy of EFW can improve about 65% from week 17 to 37 of gestation.