本文用电脑多维空间的模式识别技术,对恶性体腔积液的五项检测指标进行了分类、识别和预报。探索结果表明:红细胞数(万/mm~3),蛋白总量(g％)和粘蛋白含量(mg％),对区分癌性腹水和非癌性腹水,可作为三个主要特征值;红细胞数和蛋白总量,对区分癌性胸水和非癌性胸水,可作为二个主要特征值。分类效果明显,约90％预报结果与临床诊断基本相符。 In this paper, computer-dimensional pattern recognition technology is used to classify, identify and forecast the five indicators of malignant effusion. The results of the exploration showed that the number of red blood cells (million/mm~3), total protein (g%) and mucin content (mg%) can be used as the three main eigenvalues ​​for distinguishing cancerous ascitic fluid from non-cancerous ascites; The number and total amount of protein can be used as two main eigenvalues ​​to distinguish cancerous pleural effusion from non-cancerous pleural effusion. The classification effect is obvious. About 90% of the forecast results are basically consistent with the clinical diagnosis.