Sample of features by type for the influenza review.

Note: For alphabetic and alphanumeric+ sets, features with an underscore between pairs of words came from titles. For the indexing set, features mainly came from MeSH and Emtree; an asterisk indicates a major concept. For topic model set, number of topics determined prior to training (see Methods); based on alphanumeric+ features; Kullbach-Leibler (KL) divergences from mediods for include or exclude class. SR = systematic review. For SR concepts, lexicon consists of UMLS concepts (including parent and children) in SRs and study design terms.