Improving Understandability of Explanations with a Usage of Expert Knowledge

Data analysis is one of the most important parts of data mining and machine learning tasks. In recent years, explainable artificial intelligence methods have been used very often to support this phase. However, the explanations themselves are very often difficult to understand by domain experts, who play one of the most important roles in the phase of data analysis. In this work, we proposed a procedure to combine domain knowledge with ML and XAI methods to improve the understandability of explanations. We demonstrated the feasibility of our approach on a publicly available medical dataset. We describe a procedure for obtaining intuitively interpretable information about distinguishable groups of patients and defining differences between them with the usage of clustering, rule–based encoded domain knowledge, and SHAP values.