Feature Extraction and Protocol Adaptation: All potential queries and annotations for the data elements to be analyzed should be pre-specified. However, in our experience, it is often the case that there is a need to review the protocol after the analysis has started. To consider this need systematically, we suggest approaching the feature extraction in two waves, a preliminary wave with a pre-decided subset of datasets, followed by a revision of the protocol, and a second wave with the rest of the datasets to be annotated. That is, discuss the set of features to be extracted for each data element among members of the team from a preliminary annotation of the first n datasets to finalize the protocol of annotation for the whole set. Once a curation approach is adopted (as noted in the READ-ME file), keep it consistent.