Skip to main content
. 2020 Jan 16;27(3):407–418. doi: 10.1093/jamia/ocz207

Figure 6.

Figure 6.

Entity-level performance of medExtractR after tuning and customization on (Medical Information Mart for Intensive Care III) validation sets. Results are based on 100 validation notes after tuning and customization for each drug. P, R, and F1 represent precision, recall, and F-measure (F1 score), respectively. Tuning includes note preprocessing, dictionary updates, and optimizing parameter values. Customization refers to encoding different rules within medExtractR. Tuning and customization were developed on a separate tuning set of 10 notes for each drug. Dose change results are not presented, as there were <5 mentions within each drug.