AI-based algorithms are excellent tools for processing mass data. The great advantage of AI is the high performance of its processing. If the properties of explainability and interpretability can be technically implemented, trustworthy assistance systems in medicine are highly feasible. However, one limitation of current AI systems is that they can only justify their decisions and thus make them comprehensible if an architecture is created that makes this possible. This entry presents three examples of AI architecture implementations from the fields of ECG analysis, multimodal sleep data analysis, and contactless vital signs monitoring, which can create trustworthiness in medical applications.