Intelligence artificielle et endoscopie : le meilleur des mondes ?

Artificial intelligence (AI) aims to simulate the human intelligence. It is a cognitive science which relies on neurobiology, logical and critical thinking (problem solving, deep learning, neural networks), computing sciences (calculation, internet), and on databases. Big data exploitation (epidemiology, predictive medicine) and "signals" analysis (EKG, EEG, imaging, pathology, dermatology, ophthalmology...) were the first successful application of AI in healthcare, followed by government approval. AI has a vast spectrum of potential applications in digestive endoscopy as well. AI can be used for screening, diagnosis, characterization, treatment, and prognosis evaluation, in a wide array of procedures. The quantity of published work in this field is thriving. Computer-assisted detection and characterization of colonic polyps for instance, were amongst the first successful applications of AI, and should be commercially available shortly. The automated reading of a capsule endoscopy, based on a network of machine learning systems, is also very demonstrative of what AI will be able to accomplish in the next future. It is believed that AI will significantly improve diagnostic performances and thus the quality of care. Today, endoscopists should not only promote this technological revolution, but also address new issues in the field of AI, regarding the respective roles of physicians (focused on ethics and patient-relations) and AI-machines (assistants vs autonomous), as well as responsibility (physicians vs. manufacturing companies), and reimbursement (physician vs manufacturing companies). [ABSTRACT FROM AUTHOR]