ABSTRACT
Prediction scoring is naturally subject-centered; that is, the goal is to predict the outcome for an individual. An example of prediction scoring is the ISS score utilized in the trauma setting. Assessment scoring explores either the population or individual subjects. The scale used for any scoring system is almost irrelevant, and any system can be normalized into 0–1 scale. The natural requirements for symmetry and cumulativeness are usually met “by default” in all scoring systems. Asymptotic requirements, as a rule, are met for subject-driven scoring, and in the majority of cases it is broken for population assessment. The major problem of population-driven scoring is the inability to account for interceptions of the observed dichotomies. Ranking tests demonstrated full agreement with the blinded review and sensitivity was almost independent from the applied scores, which indicates the exceptional ability of the scoring system to assign proper ranks to every subject.
