We are not ready yet: limitations of state-of-the-art disease named entity recognizers.

We argue that there is a need for larger annotated data sets for training and testing. Therefore, we foresee the curation of further data sets and, moreover, the investigation of continual learning processes for machine learning-based models.