应用人工神经网络建立肺癌血清肿瘤标记物诊断模型的研究

目的 检测肺癌患者血清中的肺癌肿瘤标记物，应用人工神经网络建立肺癌血清肿瘤标记物诊断模型。方法 应用酶联免疫吸附法（ELISA）分别测定86例肺癌和80例健康人血清细胞角蛋白21 1片段(CYFRA21 1)、神经元特异性烯醇化酶(NSE)、组织多肽特异性抗原(TPS)、可溶性白细胞介素2受体(sIL 2R)、癌胚抗原(CEA)、糖抗原242（CA242）、抑癌基因p53抗体共7种肿瘤标记物含量。用曲线下面积结合人工神经网络建立诊断模型，并将此诊断模型用于肺癌的诊断。结果 根据血清肿瘤标记物的测定结果，计算出每个肿瘤标记物的曲线下面积，应用人工神经网络建立肺癌血清肿瘤标记物诊断模型，该模型预测肺癌样本的诊断准确率84.1%，敏感性为86.3%，特异性94.8%。结论 本研究建立的肺癌诊断模型对肺癌的诊断具有较高的敏感性和特异性。

Diagnostic model of lung cancer established with tumor markers and artificial neural network

Objective    To establish a diagnostic model for lung cancer with artificial neural network(ANN) and tumor markers. Methods    Serum levels of CYFRA21-1、NSE、TPS、sIL 2R、CEA、CA242 and p53-Ab were detected in 86 lung cancer patients and 80 healthy subjects. The tumor markers were evaluated by area under curves anda diagnostic model with artificial neural network was further established. Results    The levels of CYFRA-211、NSE、TPS、SIL-2R、CEA、CA242 and p53-Ab were detected; The diagnostic mode of lung cancer was built with artificial neural network. Accuracy, sensitivity and specificity of this model were84.1%,  86.3% and 94.8%, respectively, in the diagnosis of lung cancer. Conclusion     ANN model of tumor markers has a high sensitivity and specificity in diagnosis of lung cancer.