A Corpus for Detecting High-Context Medical Conditions in Intensive Care Patient Notes Focusing on Frequently Readmitted Patients

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 1362–1367 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 1362 A Corpus for Detecting High-Context Medical Conditions in Intensive Care Patient Notes Focusing on Frequently Readmitted Patients Edward T. Moseley1,2, Joy T. Wu1,3, Jonathan Welt1,4, John Foote Jr1,5, Patrick D. Tyler1,6, David W. Grant7, Eric T. Carlson8, Sebastian Gehrmann1,9, Franck Dernoncourt1,10,11, Leo Anthony Celi1,11 1 MIT Critical Data, Laboratory for Computational Physiology, Cambridge, MA, USA 2 College of Science and Mathematics, University of Massachusetts, Boston, MA, USA 3 Harvard T.H. Chan School of Public Health, Cambridge, MA, USA 4 Wellman Center for Photomedicine, Massachusetts General Hospital, Boston, MA, USA 5 Tufts University School of Medicine, Cambridge, MA, USA 6 Department of Internal Medicine, Beth Israel Deaconess Medical Center, Boston, MA, USA 7 Department of Surgery, Division of Plastic and Reconstructive Surgery, Washington University School of Medicine, St. Louis, MO, USA 8 Philips Research North America, Cambridge, MA, USA 9 Harvard SEAS, Harvard University, Cambridge, MA, USA 10 Adobe Research, San Jose, CA, USA 11 Massachusetts Institute of Technology, Cambridge, MA, USA etmoseley@yahoo.com, joytywu@gmail.com, jonathanwelt@gmail.com, johnfootejr@gmail.com, patrickdtyler@gmail.com, dwgrant101@gmail.com, eric@ds-do.com, gehrmann@seas.harvard.edu, franck.dernoncourt@adobe.com, lceli@bidmc.harvard.edu Abstract A crucial step within secondary analysis of electronic health records (EHRs) is to identify the patient cohort under investigation. While EHRs contain medical billing codes that aim to represent the conditions and treatments patients may have, much of the information is only present in the patient notes. Therefore, it is critical to develop robust algorithms to infer patients’ conditions and treatments from their written notes. In this paper, we introduce a dataset for patient phenotyping, a task that is defined as the identification of whether a patient has a given medical condition (also referred to as clinical indication or phenotype) based on their patient note. Nursing Progress Notes and Discharge Summaries from the Intensive Care Unit of a large tertiary care hospital were manually annotated for the presence of several high-context phenotypes relevant to treatment and risk of re-hospitalization. This dataset contains 1102 Discharge Summaries and 1000 Nursing Progress Notes. Each Discharge Summary and Progress Note has been annotated by at least two expert human annotators (one clinical researcher and one resident physician). Annotated phenotypes include treatment non-adherence, chronic pain, advanced/metastatic cancer, as well as 10 other phenotypes. This dataset can be utilized for academic and industrial research in medicine and computer science, particularly within the field of medical natural language processing. Keywords: medical NLP, patient notes, document classification Accepted as a conference paper at LREC 2020 1. Introduction and Related Work With the widespread adoption of electronic health records (EHRs), medical data are being generated and stored dig- itally in vast quantities (J. Henry and Patel, 2016). While much EHR data are structured and amenable to analysis, there appears to be limited homogeneity in data complete- ness and quality (Weiskopf and Weng, 2013), and it is esti- mated that the majority of healthcare data are being gener- ated in …

Content truncated to 500 words