Mathematical analysis of the API enteric 20 profile register using a computer diagnostic model

Abstract

Results for 21 biochemical tests using the API-20 Enteric kit were obtained from the manufacturer's files for 27,820 bacterial isolates. These isolates were identified by the API Profile Register and also by a computer diagnostic model which estimates the relative likelihoods of various identifications. The computer confirmed the identification in the API Profile Register for 99.36% of the isolates. The manufacturer has reviewed areas of the API Profile Register questioned by the computer analysis; a number of resulting modifications to the API Profile Register have been incorporated in an update letter. This computer model provides a convenient and powerful way to interpret a large number of test results for bacterial identification. This study also demonstrates the use of a large collection of isolates to refine the data matrix used by the diagnostic model.