Ce podcast est une ligne directrice vivante de l'American Gastroenterological Association (AGA) sur la prise en charge pharmacologique de la rectocolite hémorragique modérée à sévère. Il présente des recommandations basées sur une synthèse d'études, notamment des méta-analyses en réseau, concernant l’efficacité et la sécurité de divers traitements. Les recommandations portent sur le choix initial et subséquent des médicaments, incluant des thérapies biologiques avancées et des immunomodulateurs, ainsi que sur les stratégies thérapeutiques (thérapie descendante vs. ascendante). L'accent est mis sur la prise de décision partagée entre le patient et le praticien, tenant compte des facteurs individuels et des risques liés aux traitements. Des outils d'aide à la décision clinique sont fournis. AGA Living Clinical Practice Guideline on Pharmacological Management of Moderate-to-Severe Ulcerative Colitis Singh, Siddharth et al.Gastroenterology, Volume 167, Issue 7, 1307 - 1343 Hébergé par Ausha. Visitez ausha.co/fr/politique-de-confidentialite pour plus d'informations.

This document is a living guideline from the American Gastroenterological Association (AGA) on the pharmacological management of moderate-to-severe ulcerative colitis. It presents recommendations based on a synthesis of studies, including network meta-analyses, concerning the efficacy and safety of various treatments. Recommendations cover initial and subsequent drug selection, including advanced biologic therapies and immunomodulators, as well as therapeutic strategies (top-down vs. bottom-up therapy). The emphasis is on shared decision-making between patient and practitioner, taking into account individual factors and treatment risks. Clinical decision aids are provided. Hébergé par Ausha. Visitez ausha.co/fr/politique-de-confidentialite pour plus d'informations.

Le podcast présente des extraits de la revue NEJM AI, traitant de l'application de l'intelligence artificielle (IA) en médecine. Il explore des études sur une application mobile pour le dépistage précoce de l'autisme chez les jeunes enfants, un essai contrôlé randomisé utilisant l'IA pour accélérer le diagnostic d'infarctus du myocarde, et une comparaison des performances de GPT-4 avec celles de médecins lors d'examens officiels. Enfin, des perspectives sur l'utilisation de l'IA pour le conseil aux patients et le diagnostic de cas cliniques complexes sont offertes. Hébergé par Ausha. Visitez ausha.co/fr/politique-de-confidentialite pour plus d'informations.

1. How is AI being used to diagnose autism in young children? The AI Revolution in Medicine describes the use of an app called Sense To Know to analyze behavioral data from toddlers to predict a diagnosis of autism. The app uses videos of the child watching social and non-social movies, as well as playing a tactile game, to extract features such as head movement, gaze, eyebrow and mouth movements, and blink rate. 2. Can AI improve treatment times for patients with ST-segment elevation myocardial infarction (STEMI)? The ARISE study, described in “Artificial Intelligence-Enabled Electrocardiogram Alert Intervention and All-Cause Mortality”, assessed the impact of an AI-ECG system on the time elapsed between a patient's arrival at hospital and primary percutaneous coronary intervention (PCI). The study revealed that the AI-ECG intervention significantly reduced door-to-balloon time in STEMI patients, from 83.6 minutes to 78 minutes. This suggests that AI can play a crucial role in accelerating critical care for cardiac patients. 3. How does the performance of large language models (LLMs) compare with that of doctors during medical examinations? A study by Katz et al, entitled “GPT versus Resident Physicians - A Benchmark Based on Official Board Scores”, compared the performance of GPT-3.5 and GPT-4 models with that of resident physicians on official Israeli medical board examinations. GPT-4 outperformed doctors in some specialties, notably psychiatry, while its performance was weaker in others, such as pediatrics and obstetrics and gynecology. GPT-3.5 generally performed less well than GPT-4 and most physicians. 4. Are LLMs able to diagnose diseases from clinical cases? Svenstrup et al. explored the ability of the GPT-4 model to diagnose diseases from clinical cases published in online medical journals. The study revealed that GPT-4 could correctly diagnose 57% of cases, outperforming medical journal readers (36%). This suggests that LLMs could serve as valuable tools to help doctors make diagnoses. 5. What are the current limitations of LLMs in the medical field? Despite their impressive performance, LLMs still have significant limitations in the medical field. They cannot interpret medical images, which limits their usefulness in certain specialties. What's more, their knowledge is based on the data they have been trained on, which can lead to bias and errors. Finally, they lack the clinical judgment and practical experience of human doctors. 6. Can AI replace doctors? Although AI has the potential to revolutionize medicine, it is unlikely to completely replace doctors in the near future. AI can automate certain tasks and provide valuable information, but doctors are still needed to interpret results, make complex clinical decisions and provide empathetic patient care. 7. What are the ethical implications of using AI in medicine? The use of AI in medicine raises important ethical issues, including data privacy, algorithmic bias, liability and trust. It is crucial to develop clear ethical guidelines to ensure that AI is used responsibly and fairly in medicine. 8. What is the future of AI in medicine? AI is set to play an increasingly important role in medicine in the future. Continued advances in machine learning and natural language processing are likely to lead to even more powerful and sophisticated AI systems, capable of improving diagnosis, treatment and patient care. Hébergé par Ausha. Visitez ausha.co/fr/politique-de-confidentialite pour plus d'informations.