6533b860fe1ef96bd12c3876

RESEARCH PRODUCT

On the Construction of Optimum Categories in Biomedical Data Recognition Problems

U. WellnerK. BroddaC. F. Hess

subject

Pattern vectorComputer sciencebusiness.industryQuadratic classifierMachine learningcomputer.software_genreField (computer science)Task (project management)Biomedical dataClassification ruleParametric methodsArtificial intelligencebusinesscomputer

description

The recognition of patterns within sets of biomedical data involves the following problems: a) Proper recording of the data to be used b) Extraction of suitable features c) Choice of categories or classes which are relevant to the medical decision task d) Estimation of the underlying distributions in the case of using parametric methods e) Choice of an adequate classification rule Whereas a lot of theories and procedures exists for most of these steps — particularly in the field of computer-aided differential diagnosis of electrocardiograms (ECG) (see [6]) — there has been only rare considerations on the problems of definition of appropriate categories.

yearjournalcountryeditionlanguage
1979-01-01
https://doi.org/10.1007/978-3-642-93120-8_69