神经网络模型诊断胃肠道早癌价值的Meta分析

The diagnostic value of neural network models for early gastrointestinal cancer:a Meta-analysis

目的:系统评价神经网络模型诊断胃肠道早癌的价值。方法:计算机检索PubMed、EMbase、Web of Science、the Cochrane Library以及知网、万方、维普等数据库,搜集神经网络模型诊断胃肠道早癌的诊断性研究,检索时限为2010年1月至2019年10月。2名研究者按纳入与排除标准独立筛选文献、提取资料、运用QUADAS-2工具评价纳入研究的偏倚风险后,采用META-DISC 1.4和STATA 15.0软件进行Meta分析,计算合并敏感度、特异度、阳性似然比、阴性似然比、诊断比值比,绘制总受试者工作特征曲线并计算曲线下面积。结果:最终纳入了21篇文献的25个研究,包括13711张图像。Meta分析结果显示,神经网络模型诊断胃肠道早癌的合并敏感度为0.95(95%CI为0.94~0.95),合并特异度为0.90(95%CI为0.89~0.90),合并阳性似然比为7.00(95%CI为2.98~16.42),合并阴性似然比为0.09(95%CI为0.05~0.17),合并诊断比值比为78.23(95%CI为33.71~181.52)。总受试者工作特征曲线下面积为0.97。结论:神经网络模型对胃肠道早癌的诊断具有明确价值,值得进一步研究及推广。

Objective:To systematically evaluate the performance of neural network models in the diagnosis of early gastrointestinal cancer.Methods:Studies on the diagnosis of early gastrointestinal cancer using neural network models published between January 2010 and October 2019 were acquired from databases such as PubMed,EMbase,Web of Science,the Cochrane Library,CNKI,WanFang Data and VIP.Two researchers independently screened the studies,extracted data,and assessed the risk of bias for incorporating each study using the QUADAS-2 tool.Meta-analysis was then performed with META-DISC 1.4 and STATA 15.0.The pooled sensitivity,specificity,positive likelihood ratio,negative likelihood ratio and the diagnostic odds ratio were calculated,and the summary receiver operating characteristic curve was plotted and the area under the curve was calculated.Results:A total of 25 studies from 21 articles were selected,featuring 13,711 images.The results of Meta-analysis showed that,the pooled sensitivity was 0.95(95%CI:0.94-0.95),the pooled specificity was 0.90(95%CI:0.89-0.90),the pooled positive likelihood ratio was 7.00(95%CI:2.98-16.42),the pooled negative likelihood ratio was 0.09(95%CI:0.05-0.17),the pooled diagnosis odds ratio was 78.23(95%CI:33.71-181.52),and the area under the summary receiver operating characteristic curve was 0.97.Conclusion:Neural network models have significant value in the diagnosis of early gastrointestinal cancer and are worth further studying and popularizing.