AI_Colonoscopy_for_Fellows

Efficacy of Artificial Intelligence Assisted Colonoscopy in Improving Adenoma Detection Rates for Gastroenterology Fellows

The goal of this study is to look at how good artificial intelligence (AI) assisted colonoscopy is compared to conventional colonoscopy in detecting polyps when used by gastroenterology fellows. The main question the study aims to answer are:

Does AI assisted colonoscopy improve adenoma detection rate for gastroenterology fellows in different levels of training? Does AI assisted colonoscopy improve other quality indicators such as sessile serrated lesion detection rate withdrawal time? This research will compare these quality measures for gastroenterology fellows in their first, second, and third year of training. The gastroenterology fellows performing colonoscopies will be randomized to either AI assisted colonoscopy or conventional colonoscopy for each case. The nurse in the room will then turn on the AI device if needed and the rest of the colonoscopy will be completed as normal. The procedure notes will indicated whether AI device was used or not during the procedure.

The proposed study is a randomized controlled trial to evaluate the efficacy of AI assisted colonoscopies in improving colonoscopy quality indices for gastroenterology fellows.

The subjects for this study will be the gastroenterology fellows in their first, second or third year of training. Each fellow during their VA endoscopy rotation will be randomized to perform colonoscopies using AI or following conventional methods. If randomized to AI, the endoscopy technician in the procedural room will be responsible for turning on the GI Genius software (Medtronic AI device) during the procedure. The fellow will identify whether AI was used in the procedural note documented under Provation MD software. Otherwise, all colonoscopies will be performed in the standard fashion under moderate sedation or monitored anesthesia care. Procedures will not be affected by study participation. All procedures will be performed at the Oklahoma City VA Medical Center during normal procedural times. Information such as indication for the procedure and quality indices mentioned above (such as ADR, SSLDR, withdrawal time) will be obtained from chart review and calculated using data from Provation MD retrospectively, at a later time. Other information such as patient's medical history and family history as it relates to the colonoscopy procedure will also be collected retrospectively from CPRS chart records. This will be used to exclude patients undergoing colonoscopy secondary to genetic cancer syndromes (ex: polyposis), inflammatory bowel disease, for therapy of known neoplasms or undergoing non-colonoscopy screening tests (fecal immunochemical test, CT colonography) from this study. No identifiable patient information will be saved in any format. In addition, data such as polyp size, location of polyp and histological information will be collected from CPRS charts.

Inclusion Criteria:

• All adult gastroenterology fellows at the University of Oklahoma, except for the fellow conducting research will be included in the study.

Exclusion Criteria:

• All patients undergoing colonoscopy for genetic cancer syndromes (ex: polyposis), inflammatory bowel disease, for therapy of known neoplasms or undergoing non-colonoscopy screening tests (fecal immunochemical test, CT colonography) will be excluded from this study.