Two-level classification schemes in medical diagnostics.

Abstract

A two-level classifier for medical applications is considered. Such classifiers are expected to yield a more precise result than classical one-level classifiers. The underlying idea for two-level classification is supported by the routine practice of physicians to confirm the diagnosis by several data-driven inferences. An overview of the types of the two-level classifiers is presented. The competitive two-level classifier is emphasized. Three examples with real clinical data are presented from the fields of cardiology, aviation medicine, and neonatology.