The project addresses the need for a better understanding of the decisions made by artificial intelligence (AI) models in healthcare. Traditional approaches based on deep learning often function as "black boxes," making them difficult to interpret and reducing user confidence. The goal of the project is to develop explainable AI interfaces that will enable healthcare professionals to communicate effectively with machine learning models using conversational methods.

Instead of complex visualizations that require technical knowledge, the project focuses on developing conversational interfaces that explain model decisions using natural language. This includes both local (individual) and global (general) explanations, as well as the incorporation of the latest scientific findings for a better understanding and acceptance of model results.

The project has three key objectives:

• Developing conversational explanation methods that provide insight into how models work through interaction in natural language.

• Expanding to so-called "broad" explanations (broad-XAI), which include contextual and global information and relevant scientific literature.

• Personalization of explanations tailored to the characteristics of individual users (e.g., level of knowledge, visual memory) in order to increase understanding and trust.

The project addresses important challenges in the use of AI in healthcare and contributes to the development of user-friendly, trustworthy, explainable AI systems.