Computer analysis as an aid in identification of mycobacteria

One hundred eighty-four isolates representing 23 species of mycobacteria were identified using computer-assisted analysis. All isolates were examined using a standard series of 12 biochemical tests. These tests were selected because of their reproducibility and ease of performance in the laboratory. Data from these tests were analyzed by a computer that had been previously programmed to process the information and make a species determination. Identifications from the probability model were compared to identifications from conventional methods. There was 96.7% agreement between the 2 methods. The computer-assisted data analysis for identification provides increased accuracy over conventional methods because a statistical probability is applied. It also requires less time. Differences in computer data between mycobacterial species are discussed.