Handing colonoscopy prep education to an AI app — a 140-patient randomized trial against conventional nursing instruction — a clinical trial (ClinicalTrials.gov)

1When the quality of an examination rides on the quality of an explanation

Colonoscopy is central to finding colorectal cancer, and its accuracy depends heavily on whether bowel preparation the day before went well. If the bowel is not clean enough, polyps and adenomas stay hidden and the examination may have to be repeated. Preparation involves detailed steps — how to take the laxative, what to avoid eating — and the outcome turns on whether the patient understands why and follows through.

That explanation has traditionally been delivered by nurses in writing and speech, but under workload pressure and staffing shortages it is hard to give enough time to, which has been linked to poor comprehension and adherence.

2Paper and speech, or an app and a dialogue

The AI here is aimed not at diagnosis or treatment but at the work of explaining things to patients. Debate about medical AI gathers around imaging and prediction models, yet what decides whether an examination succeeds is often whether the explanation landed — and that is the first thing cut when hands are short. Of the 803 AI-related clinical trials this site holds as of 2026-09-02, 48 relate to nursing.

Alongside diagnostic support, a steady share of trials address the workload of the ward itself.

3Measuring the subjective and the objective together

The choice of endpoints is worth noting. Bowel cleanliness is scored objectively on the Aronchick Scale, an established measure, while satisfaction with nursing education and cognition about the preparation are captured separately. Cognition is measured before and after the intervention, so what changed and by how much can be seen directly.

Reporting that an app felt pleasant to use says nothing about the quality of the examination; conversely, cleanliness alone does not explain why anything improved. Only side by side do they show what changing the method of explanation meant.

4A trial that has finished

The trial began in August 2025 and completed in December of the same year. Of the 803 AI-related clinical trials this site holds as of 2026-09-02, 179 are in a completed state, with most still recruiting or not yet started.

Completion means results may exist, but what is set out here is the design and objective recorded in the registry — it does not mean app-based education was shown to be better than conventional nursing instruction. The study was also run at a single institution in southern Taiwan, with the local cultural and clinical context explicitly built into the design, which is part of the frame for reading any result.