Diagnostic test accuracy: enter 2×2 counts per study (TP, FP, FN, TN). Plots each study in ROC space, fits the Moses SROC curve, and flags threshold effect via the Spearman correlation between logit(Se) and logit(1−Sp).
Study data
ROC space + Moses SROC
Moses SROC: unweighted OLS of D = logit(Se) − logit(1−Sp) on S = logit(Se) + logit(1−Sp) (Moses et al., Stat Med 1993). The unweighted variant is the more conservative form and is what metafor::reitsma falls back to when bivariate convergence fails. Bivariate (Reitsma 2005) / HSROC (Rutter & Gatsonis 2001) require iterative ML — out of scope for this app.