As (explainable) artificial intelligence becomes increasingly integrated into high-stakes domains like healthcare, it is paramount to understand what makes explanations effective and convincing. In this work, we propose an approach that integrates argumentation schemes and Bayesian networks. The goal is to enhance the transparency and interpretability of medical diagnostic decision-making based on machine learning models. We design a novel argumentation scheme based on Walton’s abductive inference scheme that captures the reasoning process underlying medical diagnoses. The proposed scheme functions as a structured explanation template, which is instantiated with the conditional probabilities derived from a Bayesian network. These conditional probabilities are turned into statistical evidence that can support or challenge a conclusion made explicit in the argumentative scheme, thereby providing a robust and transparent basis for decision-making. The resulting explanations were evaluated in a user study by medical experts, who assessed their value and answered targeted questions about their usefulness and clarity. We present the results of this user study and provide directions for future work.