最优生化指标检测模型在4种疾病中的应用价值初探  

Primary exploration the application value of optimal biochemical items detection model in four diseases

作　　者：朱喜丹[1] 田刚[1] 蔡美珠[1] 

机构地区：[1]泸州医学院附属医院检验科,四川646000

出　　处：《国际检验医学杂志》2013年第4期437-438,共2页International Journal of Laboratory Medicine

摘　　要：目的建立4种疾病最优生化指标的检测模型,探讨不同模型在疾病诊断中的应用价值。方法以肺癌、乳腺癌,直肠癌和淋巴瘤患者为研究对象(113例),采用受试者工作特征(ROC)曲线和人工神经网络(ANN)建立诊断模型。结果 ANN模型中31项生化检测项目在肺癌、乳腺癌,直肠癌和淋巴瘤中的ROC曲线下面积(AUC)分别为0.777,0.848,0.827和0.733,而优化检测项目后4种疾病的AUC分别为0.869,0.949,0.859,0.947。结论不同疾病最优生化检测模型的建立提高了疾病诊断的准确,有助于减轻患者医疗费用。Objective A supervised biochemical detection model for four different diseases has been established to evaluate its possible application value for diagnosis of disease. Methods The model of biochemical items of 113 samples of carcinoma of the lungs,breast cancer,carcinoma of the rectum and lymphoma were established by receiver operating characteristic（ROC） curve and artificial natural neural network（ANN）. Results The area under the ROC（AUC） of 31 detection items in carcinoma of the lungs, breast cancer,carcinoma of the rectum and lymphoma were 0. 777,0. 848,0. 827 and 0. 733, respectively. The AUC of optimal model of biochemical items in four diseases were 0. 869,0. 949,0. 859 and 0. 947, respectively. Conclusion The established optimal models of biochemical items in different diseases contribute to the enhancement of veracity and alleviation of hospitalization cost of patients.

关 键 词：肺癌 乳腺癌 直肠癌 淋巴瘤 ROC曲线 

分 类 号：R446.1[医药卫生—诊断学]