Effect of atypical antibiotic resistance on microorganism identification by pattern recognition.

摘要: We classified microorganisms from the clinical laboratory by using information provided Gram stain and antibiotic sensitivity profiles obtained with Bauer-Kirby technique. Approximately 4,000 microorganisms, routinely identified tested for sensitivities in a large hospital microbiology laboratory, were used as data set several pattern recognition classification methods: K--nearest-neighbor analysis, statistical isolinear multicomponent Bayesian inference, linear discriminant analysis. analysis yielded highest prospective accuracy gram-negative organisms, 90%. When those organisms displaying an atypical resistance excluded data, improved to 95%. These results are inferior currently accepted biochemical identification methods. Microorganisms patterns likely be misidentified common enough (17% of our isolates) limit feasibility routine their sensitivities.