Fig. 2
From: Measuring the impact of screening automation on meta-analyses of diagnostic test accuracy
Example of effort/loss curve for a single meta-analysis using randomized order. The evolution of the sensitivity and specificity estimates for one diagnostic test “CD008803 1 GDx: inferior average” (n=48), where the candidate studies are screened in arbitrary order. The x-axis measures the number of screened studies (effort) and the the y-axis measures the summary estimates at the 25%, 50%, and 75% percentiles over 400 simulated screenings using arbitrary (pseudorandom) order. We also plot the difference to the “true” values (bottom)