BACKGROUND AND AIMUpper GI endoscopy (UGIE) identifies diverse findings in GI bleeding (GIB) patients. Gastric antral vascular ectasia (GAVE), a notable GIB cause, lacks a comprehensive classification. Besides, a detailed photographic data bank is crucial for training novice endoscopists and developing artificial intelligence (AI)-driven neural networks. This study aims to compile a databank of UGIE findings in GIB patients and propose a novel GAVE classification.MATERIALS AND METHODSA retrospective analysis of prospectively collected data was manually performed. Case records were evaluated, including video and photographic UGIE findings and indications for GIB patients. Fully and partially completed records were included; grossly incomplete ones were excluded. Data on common findings, indications, laboratory parameters, and risk factors were analyzed and classified. A novel classification of GAVE was developed.RESULTSOf 1821 records, 430 were analyzed. The mean patient age was 44.5 ± 16.8 years (range 10-85), with males comprising 251 (58.4%). The primary indication for UGIE was anemia (388, 90.2%), followed by hematemesis (57, 13.3%). The most frequent endoscopic findings were varices (139, 32.3%) and portal hypertensive gastropathy (131, 30.5%). GAVE was present in 22 (5.1%). Key risk factors included non-vegetarian diet (102, 23.7%) and alcohol consumption (101, 23.5%).CONCLUSIONSThis study highlights the most common indications, findings, and risk factors for UGIE in GIB patients, such as anemia, varices, and non-vegetarian diet, respectively. It provides a novel classification of GAVE and establishes a resourceful photographic data bank for beginner training and AI applications, enhancing diagnostic and therapeutic approaches in GIB.