Machine-learning (ML) studies can inflate their conclusions due to methodological biases or errors [1]. Coupled with fast-paced innovation in the world of artificial intelligence (AI), this can lead to excessive confidence in the capacities of AI and ML [2]. This is the case in clinical AI: Maleki et al. showed how three major methodological pitfalls (violation of the independence assumption, model evaluation with inappropriate metrics or baselines, and batch effect) led to unrealistic performan…