User perceptions of an artificial intelligence-based computer-aided detection system in colonoscopy: a single-center exploratory survey study

The study provides a before-and-after description of endoscopy unit staff perceptions around real-world implementation of an AI-CADe system, with positive responses regarding interest in AI-CADe, perceived adenoma detection rate improvement, expected lesion removal, patient satisfaction, and procedural satisfaction reported in both survey phases. Prior discussion of AI-CADe has often centered on diagnostic performance, whereas this study shifts attention to user acceptance, usability, and workflow integration, using a repeated cross-sectional design at two time points before and one month after installation. Single-center, repeated cross-sectional descriptive survey with 29 pre-installation and 25 post-installation participants completing anonymous online questionnaires; analyses were descriptive and exploratory.

While affirming the value of AI-CADe, staff also reported concerns about overdetection, unnecessary biopsies, and increased dependence on AI, and open-ended feedback included both positive comments on lesion recognition and procedural support and limitations related to repeated alerts, false-positive alarms, delayed detection, and system responsiveness. The study presents positive perceptions alongside specific concerns, indicating that user acceptance is not a single dimension but is intertwined with issues of false-positive detection, workflow integration, and AI dependence. Based on descriptive results from survey items on concerns regarding false-positive detection, dependence on AI, satisfaction, and open-ended feedback.

Among factors influencing acceptance and continued use, accuracy and sensitivity were the most frequently selected, followed by cost and false-positive rates. The study ranks adoption factors by frequency of selection, offering direct user-side information on which system attributes endoscopy unit staff value most. Derived from descriptive statistics on factors influencing acceptance and continued use across the pre- and post-installation surveys.

The study is suited to understanding how endoscopy unit staff at a single national cancer center perceived an AI-CADe system before and after installation, and can inform user-centered optimization, training, and workflow design; its conclusions are positioned at the level of user perception rather than diagnostic performance or patient outcomes.

As a single-center exploratory survey, the sample size is limited and the analyses are descriptive; the post-installation survey was conducted one month after system installation, so changes in perception after longer-term use remain to be observed. In addition, the currently loaded text is summary-level content and does not include the specific questionnaire items, frequency distributions of response options, or full details of open-ended feedback, which would affect further interpretation of the strength of perceptions and differences between phases.