Fig. 1From: Building a tobacco user registry by extracting multiple smoking behaviors from clinical notesVisualization of information extraction for smoking status classification. Visualization of the process used by the smoking status algorithm. Notes were manually annotated for smoking behavior concepts in 756 local records. Many of these notes were semi-structured and contained clearly defined sections. Smoking behaviors were most often found in the “social history” and impression sections, followed by the “impressions” and “health summary” sectionsBack to article page