Amber Stubbs

Software

MAE - Multipurpose Annotation Environment: lightweight, easy-to-use software written in Java for annotating written texts. Runs on most OSes (Linux, OS X, Windows XP, 7, and 8; not yet tested on Windows 10). This software was used to annotate the 2012 and 2014 i2b2 shared task corpora.

MAI - Multi-document Adjudication Interface: lightweight adjudication software that takes as input files from MAE and allows easy adjudication of different annotations to create a gold standard. Works on any systems that MAE works on.

Shared tasks

2014 i2b2/UTHealth shared task clinical NLP - featured four tracks related to longitudinal clinical narratives. Created a new de-identified corpus of 1,304 medical records representing 296 patients.

2016 CEGS N-GRID Shared-Tasks and Workshop on Challenges in Natural Language Processing for Clinical Data . Featured three tracks: de-identifiaction track (with "sight unseen" and traditional sub-tasks), Research Domain Criteria (RDoC) identification, and a novel data use track.

2018 n2c2 Natural Language Processing shared task on clinical data. This shared task featured two tracks:

  1. Cohort Selection for Clinical Trials
  2. Adverse Drug Events and Medication Extraction in EHRs

Guest Editorships

Journal of the American Medical Informatics Association, Special Issue: Special Focus Issue on 2018 n2c2 Shared-Task and Workshop on Adverse Drug Event Extraction In progress.

Journal of the American Medical Informatics Association, Special Issue: Special Focus Issue on Cohort Selection for Clinical Trials In progress.

Journal of Biomedical Informatics, Supplement: 2016 i2b2 Natural Language Processing Challenge in Clinical Data. December 2017.

Journal of Biomedical Informatics, Supplement: 2014 i2b2 Natural Language Processing Challenge in Clinical Data. December 2015.

Journal of Biomedical Informatics Volume 46, Supplement: 2012 i2b2 Natural Language Processing Challenge on Temporal Relations in Clinical Data. December 2013.