MODELING BOOLEAN DECISION RULES APPLIED TO MULTIPLE-OBSERVER DECISIONSTRATEGIES

A model that derives multiple-observer decision strategy ROC curves fo r boolean decision rules applied to binary decisions of two or three o bservers is presented. It is assumed that covert decision variables co nsistent with ROC models of observer performance underlie decisions an d that readers' decision criteria are in a fixed relationship. The spe cific parameters of individual ROC curves and the correlational struct ure that describes interobserver agreement have dramatic effects upon the relative benefits to be derived from different boolean strategies, A common strategy employed in clinical practice, in which the overall decision is positive if any observer makes a positive decision, is mo st effective when the readers are of similar ability, when they adopt similar decision criteria, when interreader agreement is greater for n egative than for positive cases, and when the individual ROC slope is much less than 1.0. Different multiple-observer decision strategies ca n be evaluated using the model equations. A bootstrap method for testi ng model-associated hypotheses is described.