Abstract Background Capsule endoscopy (CE), is a minimally invasive diagnostic tool critical for evaluating small bowel conditions, particularly Crohn's disease. Detecting ulcers and erosions is essential for assessing disease activity, guiding treatment decisions, and monitoring therapeutic responses. While CE revolutionized lesion detection, its manual interpretation is time-consuming and prone to variability. Artificial intelligence (AI) tools, including convolutional neural networks (CNNs), have emerged as transformative solutions, improving detection accuracy, reducing variability, and enabling faster, more reliable lesion identification. This study aimed to develop and validate an AI model capable of detecting and differentiating small bowel ulcers and erosions across multiple CE devices and clinical centers. Methods A multicenter, prospective study was conducted from January 2021 to April 2024, involving centers in Europe (Portugal and Spain) and the USA. The study utilized two CE devices (PillCamSB3 and Olympus EC-10) and analyzed 137 anonymized CE exams. The AI-assisted readings generated by the deep learning model were compared with standard-of-care (SoC) readings, using expert board consensus as the gold standard. Comparisons between SoC reading and AI-assisted reading were assessd by standard metrics (sensitivity, specificity, PPV, NPV, AUROC). Results Ulcers and erosions were detected in 56 patients (40.9%) during expert board review. SoC had 60.7% sensitivity, 98.8% specificity, 97.1% PPV, 78.4% NPV and 83.2% accuracy for detection of ulcers and erosions. AI-assisted reading detected ulcers and erosions with 94.6% sensitivity, 80.2% specificity, 76.8% PPV, 95-6% NPV and 86.1% accuracy. The AI-assisted model diagnosis was non-inferior (p<0.001) and superior (p<0.001) to conventional SoC diagnosis for detection of ulcers and erosions. The AI model identified 68 lesions compared to 35 detected by SoC demonstrating consistent performance across different CE devices and centers. These results were accompanied by a mean reading time of 239 seconds (SD 138 seconds) per exam with AI-assisted reading. Conclusion The AI-assisted model achieved superior diagnostic performance compared to SoC, with higher sensitivity (94.6% vs. 60.7%) and accuracy (86.1% vs. 83.2%), with a significant decrease in reading time.This landmark study represents the first interoperable, multicenter validation of an AI model capable of detecting and differentiating ulcers and erosions using data from multiple CE devices. By ensuring compatibility and high diagnostic performance across diverse settings, this technology is set to transform endoscopic practice and clinical management in IBD. References Piccirelli, S. et al. New Generation Express View: An Artificial Intelligence Software Effectively Reduces Capsule Endoscopy Reading Times. Diagnostics (Basel) 12 (2022). https://doi.org:10.3390/dia… Mascarenhas, M., Afonso, J., Andrade, P., Cardoso, H. & Macedo, G. Artificial intelligence and capsule endoscopy: unravelling the future. Ann Gastroenterol 34, 300-309 (2021). https://doi.org:10.20524/ao… Cardoso P, Mascarenhas M, Afonso J, et al. Deep learning and minimally invasive inflammatory activity assessment: a proof-of-concept study for development and score correlation of a panendoscopy convolutional network. Therap Adv Gastroenterol. 2024;17:17562848241251569. Published 2024 May 27. doi:10.1177/17562848241251569 Afonso, J. et al. Automated detection of ulcers and erosions in capsule endoscopy images using a convolutional neural network. Med Biol Eng Comput 60, 719-725 (2022). https://doi.org:10.1007/s11…