Efficiency Index for Binary Classifiers: Concept, Extension, and Application

Many metrics exist for the evaluation of binary classifiers, all with their particular advantages and shortcomings. Recently, an “Efficiency Index” (EI) classifiers has been proposed, based on consistency (or matching) contradiction mismatching) outcomes. This metric its confidence intervals are easy to calculate from base data in a 2 × contingency table, values can be qualitatively semi-quantitatively categorised. For medical tests, which context Efficiency Index was originally it facilitates communication risk (of correct diagnosis versus misdiagnosis) both clinicians patients. Variants (balanced, unbiased) take into account disease prevalence test cut-offs have also described. The objectives current paper were firstly extend EI construct other formulations (balanced level, quality), secondly explore utility four variants when applied dataset large prospective accuracy study cognitive screening instrument. showed that balanced quality, unbiased more stringent measures.