CADDIE

CADDIE is cloud-based artificial intelligence software that analyzes endoscopic video in real-time during colonoscopy to assist gastroenterologists in detecting suspected colorectal polyps. The software highlights potential polyp regions on the video feed but does not make the final diagnosis—the endoscopist retains full responsibility for confirming or rejecting detected lesions based on their medical judgment.

Both the subject device (version 1.5.0) and predicate device (K240044, version 1.4.13) are cloud-deployed standalone software accessed via web browser that generate green bounding box markers on colonoscopy video. The primary technological difference is that the Cecum AI convenience feature now uses a different deep learning algorithm model for cecal landmark detection, while the polyp detection algorithm and all other functionality remain unchanged.

Not stated in this summary. The document references FDA Design Control requirements (21 CFR 820.30) and Quality Management System compliance but does not cite specific consensus standards such as ISO, IEC, or ASTM.

The subject device is substantially equivalent because it maintains identical indications for use, patient population, and principles of operation as the predicate. Although the Cecum AI algorithm model changed, this convenience feature's functionality remains unchanged and standalone performance testing demonstrates frame-level accuracy of 90.38% with no new safety or effectiveness concerns, using the same testing methods and acceptance criteria that supported the prior clearance.