Human-centered XAI: Developing design patterns for explanations of clinical decision support systems

Much of the research on eXplainable Artificial Intelligence (XAI) has centered providing transparency machine learning models. More recently, focus human-centered approaches to XAI increased. Yet, there is a lack practical methods and examples integration human factors into development processes AI-generated explanations that humans prove uptake for better performance. This paper presents case study an application design approach explanations. The consists three components: Domain analysis define concept & context explanations, Requirements elicitation assessment derive use cases explanation requirements, consequential Multi-modal interaction evaluation create library patterns In study, we adopt DoReMi-approach Clinical Decision Support System (CDSS) child health. requirements assessment, user with experienced paediatricians uncovered what CDSS should provide. evaluation, second tested patterns. provided first set explainable decision support system in medical diagnosis, showing how involve expert end users process develop, more or less, generic solutions general problems XAI.