A Deep Learning-based System for the Bowel Preparation Evaluation Before Colonoscopy

A Deep Learning-based System for the Bowel Preparation Evaluation Before Colonoscopy

Educational interventions provided to patients may improve colon cleansing. The aims of this study were to develop an AI platform which can help patients evaluate the adequacy of bowel cleansing without the aid of caregivers and to investigate whether there was a statistically positive correlation between the Image rating and the BBPS score.

Inadequate bowel preparation can have multiple negative effects on colonoscopy. It has been associated with significantly lower rates of detection of adenomas and advanced adenomas in two meta-analyses. An observational study revealed a threefold higher miss rate for adenomas ≥ 5 mm in size when bowel cleansing was inadequate. Early colorectal cancers might appear as very subtle mucosal lesions. To ensure detection, these lesions require careful and complete mucosal inspection and optimal bowel preparation. Studies in various international populations have found that inadequate cleansing is a factor in approximately 20%-70% of incomplete colonoscopies. A ≥ 90 % minimum standard for adequate bowel preparation has been recommended by the Quality Committee of the European Society of Gastrointestinal Endoscopy (ESGE).

The Boston Bowel Preparation Scale (BBPS) is a rating scale used to evaluate the adequacy of bowel preparation. However ，the adequacy of bowel preparation before colonoscopy has the same significance. In clinical practice, the evaluation of bowel preparation before colonoscopy is mostly performed by the patients themselves. During bowel preparation, patients typically cannot accurately judge bowel readiness after taking laxatives. Many studies showed that educational interventions provided to patients and health care personnel may improve colon cleansing. However, patients often prepare their bowel during non-business hours such as the evening or early morning. Manual guidance during the patient's bowel preparation is not available. The aims of this study were to develop an AI platform which can help patients evaluate the adequacy of bowel cleansing without the aid of caregivers and to investigate whether there was a statistically positive correlation between the Image rating and the BBPS score.

Inclusion Criteria:

• People sign an "informed consent form"
• People able to use smartphone to follow the WeChat public account

Exclusion Criteria:

• People who have contraindications for colonoscopy
• People who have known colorectal polyps
• People with history of surgery in any part of the large bowel
• People who refuse to join the study