Text Message System for the Prediction of Colonoscopy Bowel Preparation Adequacy Before Colonoscopy: An Artificial...

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Gastroenterology

Background:

• Inadequate bowel preparation affects 25% of colonoscopies, hindering colorectal cancer screening.
• Effective colonoscopy relies on optimal bowel cleansing.

Purpose of the Study:

• To develop an AI algorithm for assessing bowel preparation quality using stool photos.
• To predict inadequate bowel preparation before colonoscopy.

Main Methods:

• Patients submitted stool photos via text message.
• A binary classification machine learning algorithm analyzed images.
• Boston Bowel Preparation Scores categorized preparation quality.

Main Results:

• The AI algorithm distinguished inadequate/fair from good preparation with 78.6% PPV and 60.8% NPV.
• Predicting colonoscopy duration was less accurate (78.6% PPV, 65.5% NPV).

Conclusions:

• Patients are willing to submit stool photos for bowel preparation assessment.
• AI image analysis shows promise in predicting bowel preparation quality.
• The algorithm's accuracy in predicting colonoscopy duration needs improvement.