Computer-aided microbial identification using decision trees

This paper will demonstrate the utility of using machine learning methods t o develop more efficient microbial identification (ID) techniques. The use of computer algorithms to create new decision trees can improve efficiency and increase systematization in the field of microbiology. Preliminary resu lts indicate that decision tree algorithms can create new structures that r equire fewer tests on average to reach a positive identification of an unkn own organism. Including test time and cost factors can make further improve ments, resulting in systems that are more time-efficient and/or cost-effect ive. Machine learning techniques can also create customized ID systems for specific applications. This paper will explain the induction of decision tr ees and show examples of their use in microbial ID. (C) 1999 Elsevier Scien ce Ltd. All rights reserved.