The empirical Receiver Operating Characteritic (ROC) curve can be used to evaluate the properties of a diagnostic test from the distribution of a variable on the healthy and diseased populations. When data are available from different sources, the Minimum Averaged Mean Squared Error (MAMSE) weights can be used to adjust the relative contribution of each data source. The authors use the MAMSE weights to infer the ROC curve of a diagnostic test based on raw data from multiple studies. The proposed estimates are consistent and Monte Carlo simulations show favorable finite sample performance. The method is further illustrated in a case study where progesterone level is used to detect ectopic pregnancies and abortions from other natural causes.
Published July 2015 , 14 pages