Strategy for Decreasing Inadequate Bowel Cleansing in Colonoscopy Based on an Artificial Intelligence System: A Randomized and Controlled Study

The patient's perception of colon cleanliness prior to undergoing a colonoscopy has been studied as a predictor of colon cleanliness quality, demonstrating to be a powerful predictor of inadequate cleanliness. A convolutional neural network developed by our group, trained with photographs of rectal effluents at different moments of colon preparation, has achieved high diagnostic accuracy. Based on all this experience, the next step would be to evaluate in a randomized clinical trial whether this neural network integrated into a computer application associated with cleaning recommendations improves the colon cleanliness quality of patients compared to a control group, being the objective of this project Therefore, the main purpose of the study is to assess if a strategy based on a mobile application linked to a neural network is useful for guiding colon cleansing in a more personalized way is better than the usual care defined as regular oral and written instructions. The secondary aim will be the acceptance of this artificial intelligence device defined as the proportion of patients assigned to the intervention group that actually used the device. Consecutive outpatient patients meeting inclusion criteria and none of the exclusion criteria who have been requested to undergo colonoscopy will be included in the study and randomized to mobile artificial intelligence application or control group The intervention group will receive a response from the AI system in order to determine the quality of colon cleansing: adequate preparation or inadequate preparation. In addition, the system will issue specific recommendations based on the quality of cleansing. Patients assigned to the control group will undergo colonoscopy preparation according to standard recommendations.