Luo H, Xu G, Li C, et al. - In this multicentre, case-control, diagnostic study, researchers constructed as well as sought validation for an imaging tool—Gastrointestinal Artificial Intelligence Diagnostic System (GRAIDS)—to diagnose upper gastrointestinal cancers, via examining imaging data from clinical endoscopies. This research was performed in six hospitals in China. They collected images of consecutive participants, aged 18 years or older, who had not had a prior endoscopy. Eligible patients were those having histologically confirmed upper gastrointestinal cancer lesions (including oesophageal cancer and gastric cancer). Random assignment (8:1:1) of images from Sun Yat-sen University Cancer Center to the training and intrinsic verification datasets was done to construct GRAIDS, and to the internal validation dataset in order to assess the performance of GRAIDS. With high diagnostic accuracy, GRAIDS allowed the detection of upper gastrointestinal cancers. A sensitivity similar to that of expert endoscopists was yielded by GRAIDS. Compared with non-expert endoscopists, it displayed a superior sensitivity. Community-based hospitals could be benefitted from this system with respect to achieving an improved efficacy in upper gastrointestinal cancer diagnoses.