Bowel Preparation, Colonoscopy
Conditions
Keywords
Artificial Intelligence, Bowel Preparation, Colonoscopy, Smartphone Application, Image Recognition, Patient Education
Brief summary
This randomized controlled trial aims to evaluate the effectiveness of a multimodal artificial intelligence (AI)-assisted smartphone application in improving bowel preparation outcomes among hospitalized adults undergoing colonoscopy. A total of 140 participants will be randomly assigned in a 1:1 ratio to either an experimental group or a control group. The control group will receive conventional written and verbal nursing education, while the experimental group will receive the same standard education plus a multimodal AI-assisted application delivered through the LINE platform. The application provides structured bowel preparation education, interactive AI-based question-and-answer support, dietary image recognition, stool image analysis, and individualized feedback. Study outcomes will include bowel preparation knowledge, satisfaction with nursing education, and bowel cleansing quality assessed using the Aronchick Scale.
Detailed description
Adequate bowel preparation is essential for high-quality colonoscopy because insufficient bowel cleansing may reduce mucosal visualization and affect the detection of colorectal lesions. Conventional bowel preparation education generally relies on written materials and verbal instructions. However, patients may have difficulty understanding dietary restrictions, bowel cleansing medication instructions, and whether bowel cleansing is adequate during the preparation process. This study evaluates a multimodal AI-assisted smartphone application designed to support patients throughout the bowel preparation process. The application is integrated with the LINE platform and provides structured educational content, including bowel preparation instructions, bowel cleansing medication guidance, dietary preparation, instructional videos, and an interactive AI chatbot for real-time individualized responses. The application also incorporates two AI-assisted image-recognition functions. First, participants may submit photographs of their meals before colonoscopy. A multimodal AI model analyzes the food images and classifies the dietary pattern as a clear liquid diet, low-residue diet, or regular diet, and provides individualized dietary recommendations. Second, after taking bowel cleansing medication, participants may submit stool images. The system uses a multimodal AI model in combination with a convolutional neural network model to assess bowel cleansing status and provide feedback. Participants in the control group will receive conventional written and verbal nursing education regarding diet, bowel cleansing medication, and colonoscopy preparation. Participants in the experimental group will receive the same conventional education plus access to the multimodal AI-assisted application. Bowel preparation knowledge will be assessed before and after the intervention, satisfaction with nursing education will be evaluated after colonoscopy, and bowel cleansing quality will be assessed by the endoscopist using the Aronchick Scale.
Interventions
BEHAVIORALMultimodal AI-Assisted Bowel Preparation Education
Participants in the experimental group receive conventional bowel preparation education plus a multimodal AI-assisted application delivered through the LINE platform. The application provides structured education on bowel preparation, dietary restrictions, bowel cleansing medication, and examination procedures; interactive AI-based question-and-answer support; dietary image recognition; stool image analysis; and individualized feedback throughout the bowel preparation process. Food images are analyzed to classify dietary patterns as clear liquid, low-residue, or regular diet, while stool images are analyzed using multimodal AI and a convolutional neural network model to assess bowel cleansing status and provide feedback.
Sponsors
Chiayi Christian Hospital
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
SUPPORTIVE_CARE
Masking
SINGLE (Outcomes Assessor)
Masking description
The endoscopist who assesses bowel cleansing quality is blinded to group allocation. Questionnaire data are collected by research personnel who are unaware of the participants' group assignments. Participants and personnel delivering the intervention cannot be blinded because of the nature of the AI-assisted application.
Intervention model description
Participants are randomly assigned in a 1:1 ratio to either an experimental group receiving conventional nursing education plus a multimodal AI-assisted application or a control group receiving conventional nursing education alone. The two groups are studied concurrently without crossover.
Eligibility
Sex/Gender
ALL
Age
20 Years to No maximum
Healthy volunteers
No
Inclusion criteria
1. Hospitalized adults scheduled to undergo colonoscopy. 2. Age 20 years or older. 3. Able to read. 4. Conscious and able to communicate in Mandarin or Taiwanese. 5. Own a smartphone and have basic ability to use the LINE application.
Exclusion criteria
1. Patients receiving hemodialysis or peritoneal dialysis. 2. Patients unable to take bowel cleansing medication orally. 3. Patients unable to consume a large volume of fluids. 4. Patients with intestinal stenosis, bowel obstruction, or obstructing intestinal tumors. 5. Patients with severe active gastrointestinal bleeding (\>1500 mL/day).
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Bowel Cleansing Quality Assessed Using the Aronchick Scale | During colonoscopy, approximately 2-3 days after enrollment | Bowel cleansing quality will be assessed by the endoscopist using the Aronchick Scale, which classifies bowel preparation as Excellent, Good, Fair, Poor, or Inadequate. Excellent and Good ratings will be considered adequate bowel preparation. |
Countries
Taiwan
Outcome results
None listed