计算机模式识别法对类风湿性关节炎中医证候判断与常规临床判断的比较

根据１６３例类风湿性关节炎患者头发１４种元素的检测结果，采用计算机模式识别法中的动态聚类法在多维空间进行分类可得出五种类型，再用其识别法中的马氏距离法将其降到一维平面进行回判，然后将回判的结果与常规临床证候诊断所得出的肝肾两虚、寒热错杂、气阴两虚、湿热阻络及痰湿阻络证型的分类结果用观察一致性进行检测，其结果Ｋ（ｋａｐｐａ）＝０．７７，＞０．６。表明两种方法一致性相当可靠，并得出计算机模式识别法对中医证候的判断的敏感度为９６．６７％，特异度为９５．１５％。提示该法有可能对类风湿性关节炎中医证候作出较为客观的判断。

Comparison between Two Diagnostic Methods of Computer's Mathematic Model and Clinical Diagnosis on TCM Syndromes of Rheumatoid Arthritis

Based on the test result of 14 trace elements in hair of 163 cases of rheumatoid arthritis,we used one of the methods of the computer's mathematic model,DYNAMIC to assign the patients to five groups among multidimensional space.Then,another method,M-DEC was used to diagnose it back on one plane.The result was compared with that by the usual clinical diagnosis,also five groups were obtained, the group of Deficiency of both Liver and Kidney,syndrome of intermingled Cold with Heat,Deficiency of Qi and Yin,syndrome of Dampness-Heat stagnating in collaterals and syndrome of Phlegm-Dampness stagnating in collaterals.And Kappa was 0.77,greater than 0.6.It displayed the consistency in observation of these two methods,which was shown reliable.After that,we have detected the sensitivity of the syndrome diagnosis by computer's mathematic model method, the result was 96.67% and the specificity of that was 95.15%.So we consider this method could give us an objective judgement on TCM syndromes of rheumatoid arthritis.