The utility of artificial intelligence in gastrointestinal oncology: A systematic review of randomized controlled trials

Abstract

In the field of gastrointestinal oncology, the development of novel artificial intelligence (AI) processes may help with multiple aspects of cancer care delivery. However, a comprehensive understanding of the current utility of AI in gastrointestinal oncology is lacking. The authors conducted searches in the following databases: MEDLINE (Ovid), Embase (Ovid), and CINAHL (Cumulative Index of Nursing and Allied Health) Ultimate (EBSCO). The analysis focused on publication trends and outcomes of randomized controlled trials (RCTs) that used AI to manage gastrointestinal malignancies. From our initial search retrieval of 3730 studies, 27 RCTs (with a total of 29,895 patients) were identified that met inclusion criteria. The first RCT was published in 2019, followed by five in 2020, four in 2021, six in 2022, and 11 in 2023. Colorectal malignancies comprised the majority of the literature (23 of 27 studies; 85%), with other studies focused on gastric cancer (three of 27 studies; 11%) and hepatocellular carcinoma (one of 27 studies; 4%). Of the included RCTs, 25 (93%) had a primary outcome focused on lesion/cancer detection throughout the gastrointestinal tract using endoscopy or ultrasound, with others focused on algorithmic-based AI assistance with postoperative pain management or histologic diagnosis. Overall, 22 of 27 studies (81%) met their primary end point with a statistically significant result. In this systematic review, the authors observed a recent increase in the number of RCTs focused on AI within the field of gastrointestinal oncology and identified specific areas in which AI is being used. Findings from this work should help to inform further investigations to develop and test innovative AI uses, enhance care delivery, and improve patient outcomes.