Abstract
This comment highlights the need for tools to ensure the clinical relevance of AI-generated solutions. Doctors have medical Concilium, AI does not. AI tries to be the best on its own. AI uses formal data processing logic, and therefore may not consider the limitations of medical data. AI may “forget and ignore” data that do not fit into its own solution. A doctor can diagnose dementia in a human but how to detect formally correct but clinically wrong suggestions by AI? Explainable AI is a good step in solving the problem of clinical relevance. However, much more must be done to translate the logic of a Doctors Concilium into tools embedded in AI diagnostics.