Moser M
Phoniatrische Abteilung, Universitäts-HNO-Klinik, Erlangen.
Laryngol Rhinol Otol (Stuttg). 1988 Mar;67(3):118-22.
Pattern recognition methods based on parameters and on geometry are described briefly and the results obtained with different characteristics in 79 patients with single potential recognition and subsequent sequential testing are presented. In subjectively very safely identifiable potentials accurate results in a proportion of up to 92% are obtained; if all tests with a subjective rating are considered, one obtains a correct automatic reclassification quota of 69% for the randomly selected sample. This system can help the examining person by supplying him with decision criteria. However, without additional processing of the entire context automatic processing is not sufficiently safe. Possibilities for the use of the automatic recognition method in scientific problems are demonstrated by means of examples.
简要描述了基于参数和几何的模式识别方法,并展示了在79例具有单一电位识别及后续序贯测试的患者中,不同特征所获得的结果。在主观上非常安全可识别的电位中,可获得高达92%的准确结果;如果考虑所有具有主观评分的测试,对于随机选择的样本,自动重新分类的正确配额为69%。该系统可以通过为检查人员提供决策标准来帮助他们。然而,未经对整个背景进行额外处理,自动处理的安全性不足。通过实例展示了在科学问题中使用自动识别方法的可能性。