The authors have developed an automated computer-aided audiogram classification system which executes pattern recognition of the standard puretone air-conduction audiogram. The system is composed of a micro-processer, floppy disk drive units, graphic color display and printer. Data of various kinds of audiograms were stored in the floppy disk through the key-board. The microcomputer detected and retrieved the characteristics of plotting curve of hearing loss level at each frequency through artifical intelligence technique.
The authors calculated the mean value of hearing level at each frequency of 3000 audiograms, and few characteristics of plotting curve pattern were decided. Each audiogram patterns were divided with some segments according to these characteristics. By using regression equation technique ao each segment, the computer-aided audiogram pattern classification procedure was established.
The ability for plotting curve recognition and judgement of hearing level utilizing the computer showed no significant differences from those done by physicians. This new system can be effectively applied in screening audiometry and automated health testing.