Electronic Resource
Springer
Annals of biomedical engineering
1 (1972), S. 115-133
ISSN:
1573-9686
Source:
Springer Online Journal Archives 1860-2000
Topics:
Medicine
,
Technology
Notes:
Abstract The introduction of computer-directed secondary testing techniques into a Multiphasic Health Testing System will bring about a need for on-line classification of patients. This paper discusses one method for classifying patients by means of their responses to a medical history questionnaire. The questionnaire data are processed through a trainable pattern classifer which uses a linear discriminant function. The algorithm selected for training purposes is based upon a fixed-increment error correction scheme and is a function of a predetermined maximum likelihood ratio value. The accuracy of the pattern classifier was tested by means of published questionnaire data relating to two different disease categories. Preliminary results indicate that the pattern classifier can be trained to a high degree of accuracy (〉90%) and that the savings in computer storage space for this classification approach is significant.
Type of Medium:
Electronic Resource
URL:
http://dx.doi.org/10.1007/BF02363422
Permalink
|
Location |
Call Number |
Expected |
Availability |