We have previously developed an original method to evaluate small bowel motor function based on computer vision analysis of endoluminal images obtained by capsule endoscopy. Our aim was to demonstrate intestinal motor abnormalities in patients with functional bowel disorders by endoluminal vision analysis. Patients with functional bowel disorders (n ϭ 205) and healthy subjects (n ϭ 136) ingested the endoscopic capsule (Pillcam-SB2, Given-Imaging) after overnight fast and 45 min after gastric exit of the capsule a liquid meal (300 ml, 1 kcal/ml) was administered. Endoluminal image analysis was performed by computer vision and machine learning techniques to define the normal range and to identify clusters of abnormal function. After training the algorithm, we used 196 patients and 48 healthy subjects, completely naive, as test set. In the test set, 51 patients (26%) were detected outside the normal range (P Ͻ 0.001 vs. 3 healthy subjects) and clustered into hypo-and hyperdynamic subgroups compared with healthy subjects. Patients with hypodynamic behavior (n ϭ 38) exhibited less luminal closure sequences (41 Ϯ 2% of the recording time vs. 61 Ϯ 2%; P Ͻ 0.001) and more static sequences (38 Ϯ 3 vs. 20 Ϯ 2%; P Ͻ 0.001); in contrast, patients with hyperdynamic behavior (n ϭ 13) had an increased proportion of luminal closure sequences (73 Ϯ 4 vs. 61 Ϯ 2%; P ϭ 0.029) and more high-motion sequences (3 Ϯ 1 vs. 0.5 Ϯ 0.1%; P Ͻ 0.001). Applying an original methodology, we have developed a novel classification of functional gut disorders based on objective, physiological criteria of small bowel function.functional bowel disorders; capsule endoscopy; intestinal motility; computer vision analysis; machine learning WE HAVE DEVELOPED AN ORIGINAL method to evaluate small bowel motor function based on computer vision analysis of endoluminal images obtained by capsule endoscopy. A series of features, such as phasic luminal closures, noncontractile patterns, e.g., smooth wall or open tunnel view, amount of chyme, and motion of the walls and content, are quantitatively measured. Each subject is then defined by a combination of values derived from these various dimensions. Definition of normality in this multidimensional space is performed by machine learning techniques: providing a large series of normal cases, the program draws the hyperplane that defines the normal range and, hence, automatically learns to discriminate what is normal and what is not (24). This may constitute a particularly valuable diagnostic tool on account of the enormous amount of data obtained in a large sample of healthy individuals and the automated algorithms that eliminate the piecemeal assessment of normality for each separate feature as well as the subjectivity of the data analyst, common to many physiological gut tests.