中文摘要：

目的：建立临床诊断的数学模型。方法：构建感知器神经网络，以癌症诊断为例，对50例非癌症患者和100例癌症患者的腺苷三磷酸酶（ATP酶）和琥珀酸脱氢酶（SDH酶）活性2项指标分组进行训练和仿真诊断。结果：2个检测样本的诊断正确率分别为96％和97．33％。结论：可以用感知器神经网络建立临床的疾病诊断系统。

英文摘要：

Objective To set up a mathematical model for clinical diagnosis. Methods After setting up perceptron neural network, as an example of cancer diagnosis, the values of adenosine triphosphatase （ATPase） and succinate dehydrogenase （SDHase） of 50 healthy people and 100 cancer patients were divided into groups to he trained and simulated. Results The correct diagnosis rates of the two samples were respectively up to 96% and 97.33%. Conclusion Perceptron neural network can he used to set up clinical diagnosis system.