A Prospective, Randomized, Evaluator Blind, Parallel Study of the Efficacy of AI-driven Feces Identifying for the Bowel Preparation Prior to Colonoscopy

Bowel preparation is the process of removing all stools from the colon in order to have a medical or surgical procedure such as a colonoscopy. It is important to clean the colon of all stools, food particles and anything else that may be present for several reasons. If you are having surgery on or near the colon, having stools presented is a risk for infection and can get in the way of some procedures. There are times when bowel preparation is done purely to prevent potential complications from surgery. For example, if your bowel was nicked during surgery, the infection risk would be dramatically decreased if the bowel was empty. In the case of a colonoscopy, stool in the colon can prevent the surgeon from seeing the tissue that is being inspected and would make it very difficult to introduce the lighted scope into the rectum and colon.

Colonoscopies have been shown to help reduce the incidence of colon cancer and deaths associated with the disease. Over the last decade, public health initiatives have helped drive up the rate of screenings, yet the current rates still fall short of public health targets set forth. Concurrently, the rate of adequate preparing for screening which is associated with colonoscopy effectiveness has changed little over the last 10 years.

With rapid advancements in artificial intelligence (AI), image recognition techniques have emerged as particularly valuable tools in medical applications. Deep learning algorithms, a subfield of AI, have demonstrated impressive performance in analyzing and categorizing complex medical imaging data, enabling efficient real-time decision support in clinical settings. Recent studies have highlighted the promising application of AI-driven image recognition in evaluating stool appearance and bowel cleanliness, providing immediate feedback and personalized instructions to patients undergoing bowel preparation. Incorporating such AI system has the potential to significantly enhance patient compliance, improve bowel preparation effectiveness, and consequently, increase diagnostic accuracy and the overall clinical value of colonoscopy procedures .This study is a prospective, randomized, evaluator-blind, parallel-group clinical trial designed to evaluate the efficacy of an AI-SIS in improving bowel preparation prior to colonoscopy. Participants will be randomly assigned in a 1:1 ratio into two groups, with approximately 170 participants per group, totaling 340 participants.

Group A (Intervention group): Participants will utilize the AI-SIS, designed to provide real-time feedback on bowel cleanliness through image classification of stool into three categories (Grade A: adequate cleansing; Grade B or C: inadequate cleansing requiring further bowel preparation). In addition to the standard hospital-provided educational instructions, this group will use the AI-SIS.

Group B (Control group): Participants will undergo standard bowel preparation and receive only the standard hospital-provided educational instructions without the assistance of the AI-SIS.