Impact de l’intelligence artificielle sur l’évolution des pratiques cliniques en oncologie : focus sur les modèles de langue

Abstract

Artificial intelligence (AI) is addressing many expectations for healthcare practitioners and patients in oncology. It has the potential to deeply transform medical practices as we know them today: improving early diagnosis by analysing large quantities of medical data, refining personalised treatment plans and optimising patient follow-up. AI also makes it easier to identify new biomarkers and predict responses to therapies, reducing margins of error and speeding up clinical decisions. Among the most popular types of AI to revolutionise clinical practice are language models. In a perfect world, the integration of AI would promote more precise, personalised and efficient care, while relieving healthcare providers of tedious or repetitive tasks, allowing them to concentrate more on providing human support to patients, and all this with a low energy consumption. However, the large-scale deployment of AI currently raises fundamental questions about fairness, safety of use and how to assess the results obtained from AI longitudinally. This article explores how the many applications are evaluated for our practice (spoiler alert: they are currently limited), potential clinical benefits and challenges currently encountered when dealing with the integration of AI into routine oncology care. We will focus on language models whose development has been exploding since 2021.

Ähnliche Arbeiten