542 Gastroenterologist Sentiment Toward Artificial Intelligence (AI) in Endoscopic Practice: A Nationwide Survey

Abstract

INTRODUCTION: Application of artificial intelligence (AI) in gastroenterology has been shown to improve the performance of endoscopists in several tasks including colonic polyp detection. However, little is known about how gastroenterologists perceive AI, especially with respect to its costs, ease of use, and efficacy. This nationwide survey is amongst the first to describe how practicing gastroenterologists view this emerging technology METHODS: In this cross-sectional observational study, an online survey was sent to practicing gastroenterology physicians and fellows across the US. This included questions about physician background as well as questions about physician understanding and perception of AI. Descriptive statistics were utilized to summarize physician sentiment toward AI. Chi-square test or Fisher's exact test were utilized to assess whether background information about physicians was correlated to their sentiment towards AI. Finally, a multivariate regression model was created to determine which physician factors were most correlated to physician sentiment RESULTS: Online surveys were sent by email to 330 gastroenterologists nationwide representing a variety of practice settings and experience levels out of which 124 physicians completed the survey and were included in the analysis. 86% of physicians reported interest in AI-assisted colonoscopy and 84.7% agreed that computer assisted polyp detection (CAPD) would improve their endoscopic performance. Univariate analysis showed that physicians practicing in academic centers versus private practices (P = 0.025) and physicians with less than 15 years post-fellowship experience versus physicians with more than 15 years of post-fellowship experience (P = .006) had a greater expectation of AI improving endoscopic performance. Multivariate analysis showed that post-fellowship experience of less than 15 years was the most important factor in determining whether a physician was likely to believe that CAPD would improve endoscopic performance (OR = 4.03; P = .04). The most common concerns about implementation of AI in endoscopic procedures were cost (75.2%) and development of operator dependence (62.8%) CONCLUSION: Gastroenterologists nationwide have an interest in the application of artificial intelligence to endoscopic procedures in gastroenterology, and many believe that this technology would improve their clinical performance. Further studies are needed to determine if AI applications in gastroenterology are cost-saving and time neutral.