摘要
目的探索利用机器学习基于尿常规检查结果的数据,进行辅助筛查泌尿系统肿瘤。方法利用PTSVM算法对500位正常人和408位泌尿系统疾病患者的尿常规数据进行分析,找到其与泌尿系统恶性肿瘤的相关性。结果对于泌尿系统恶性肿瘤,通过5次交叉验证,机器学习的最优平均分类准确率达到了85. 78%。结论 PTSVM算法可以通过尿常规检查数据区分正常人和泌尿系统恶性肿瘤患者,表明该方法有望成为一种泌尿系统肿瘤辅助筛查手段。
Objective To explore an assist method for the diagnosis of urological carcinoma based on routine urianlysis by deep learning. Methods The progressive transductive support vector machine( PTSVM) was applied to distinguish the carcinoma of urinary system by analyzing the data of routine urianlysis of 500 patients of normal persons and 408 patients of urological malignancies. Results The average accuracy rate of deep learning method through 5 cross validation is 85. 78% for urological malignancies,there was no differences between SVM and ANN. Conclusions There is probability to classify normal and urinary system malignancies patients by using deep learning method on routine urianlysis.
引文
[1]张洁,王文,伏慧.尿常规检测对尿路感染快速诊断的意义[J].宁夏医学杂志,2017,39(9):838-840.
[2]周丽丽,李晶.糖尿病患者尿常规的检测结果及检测意义探究[J].中国医药指南,2016,14(18):131-132.
[3]郎需和,王增良,李丽霞,等.以尿常规检查为主的诊治方案对急性输尿管结石的作用[J].医学综述,2015(11):2103-2104.
[4]Mangasarian OL,Wild EW.Multisurface proximal support vector machine classification via generalized eigenvalues[J].IEEE Trans Pattern Anal Mach Intell,2006,28(1):69-74.
[5]Ye Q,Zhao C,Ye N,et al.Multi-weight vector projection support vector machines[J].Pattern Recog Lett,2010,31(13):2006-2011.
[6]祁亨年.支持向量机及其应用研究综述[J].计算机工程,2004,30(10):6-9.
[7]陈素根,吴小俊.改进的投影孪生支持向量机[J].电子学报,2017,45(2):408-416.
[8]Parakh P,Bhatta NK,Mishra OP,et al.Urinary screening for detection of renal abnormalities in asymptomatic school children[J].Nephrourol Mon,2012,4(3):551-555.
[9]范宏佳,汤瑾,王坚镪,等.尿常规检测在尿路感染诊断中的价值[J].检验医学,2014,29(12):1207-1211.