Abstract(s) :
(Anglais) In a society open to technological progress, Artificial Intelligence (AI) meets only with lukewarm approval. With its increasing impact in our daily life, many see AI as the promise of a better world, while others see it as a threat of human substitution. That being said, AI achieves better acceptance as regards medical research. In this context, modern machine learning gives grounds for hope of developing diagnosis support systems. To ensure these systems present a concrete clinical applicability, it is crucial to consider the appropriate paradigms of development. We present these paradigms and provide guidelines for their implementation in practice.