We appreciate Jialin Li and colleagues' comments on our recent study, which proposes use of the Gastrointestinal Artificial Intelligence Diagnostic System (GRAIDS) for endoscopic detection of upper gastrointestinal cancers. Li and colleagues questioned whether GRAIDS is equally effective across all stages of upper gastrointestinal cancers. Unfortunately, disease staging information was not collected for the past 10 years for the large-scale retrospective images obtained from the six participating centres. Nonetheless, T (tumour invasion) staging information was available in 386 patients and staging information from The American Joint Committee on Cancer was available in 428 patients in the prospective cohort. The accuracy of GRAIDS was consistently high for detecting upper gastrointestinal cancers across different stages (92·2%, 95·3%, 93·1%, and 95·4%, respectively, for T1, T2, T3, and T4 disease; and 93·6%, 92·7%, 95·3%, and 96·2%, respectively, for stage I, II, III, and IV disease).

中文翻译：

---

人工智能在上消化道癌症中的应用-作者的回复。

我们赞赏李嘉琳及其同事对我们最近的研究的评论，该评论提出使用胃肠道人工智能诊断系统（GRAIDS）进行内窥镜检查上消化道癌症。李和同事质疑GRAIDS是否在上消化道癌症的所有阶段都同样有效。不幸的是，过去十年来没有从六个参与中心获得的大规模回顾性图像收集疾病分期信息。尽管如此，在预期队列中有386例患者获得了T（肿瘤浸润）分期信息，而有428名患者获得了美国癌症联合委员会的分期信息。GRAIDS在不同阶段检测上消化道癌症的准确性一直很高（92·2％，95·3％，93·1％，对于T1，T2，T3和T4疾病，分别为95.4％；I，II，III和IV期疾病分别为93.6％，92.7％，95.3％和96·2％）。