Abstract: Techniques for extracting clinical entities from clinical documents are provided. In some embodiments, a non-transitory machine-readable medium stores a program. The program receives a clinical document partitioned into a set of sections. Each section in the set of sections includes a plurality of terms. The program further identifies a section in the set of sections. The program also determines a subset of a plurality of entity recognizers based on a classification of the section. Each entity recognizer in the plurality of entity recognizers is configured to identify terms in the section as being associated with a particular type of data. The program further sends the section to the subset of the entity recognizers for processing. The program also generates a clinical statement based on terms identified in the section by the subset of the plurality of entity recognizers.