The formulation of the ADHD-200 Global Competition refers to a three class discrimination problem: ADHD-combined, ADHD-inattentive, control. Nevertheless it is not clear what kind of evaluation metric will be applied to the predictions submitted by the applicants. The are many ways to compute a score, and a slight difference on evaluation metric might have a strong impact on the type of solution. Is it enough to maximize the arithmetic sum of correct classifications, where 1 denote the optimal performance? Do we have to consider that a false positive has a different weight than false negative? To classify a subject as ADHD combined rather than ADHD-inattentive is less critic than to classify it as control? I believe this kind of information should be delivered as soon as possible.
Best regards.