Fig. 1.

Current mechanisms to restore or increase biological interpretability. There is a negative correlation between the interpretability of the DL model and the size of the feature space. However, the integration of domain knowledge can systematically address this dimensionality issue. Domain knowledge can be distinguished between expert-level knowledge and knowledge derived from databases. The motifs for integrating domain knowledge within emerging architectures were identified: the domain knowledge which a is integrated in the design of the models, b is integrated in the input data pre-processing, c is integrated in the post-hoc analysis process