Could an augmented, institution-wide curation of clinical notes derived from an electronic health record (EHR) on full-spectrum symptoms in patients prior to their COVID-19 testing facilitate earlier diagnoses?
Bottom Line: Understanding temporal dynamics of COVID-19 patient symptoms could provide fine-grained resolution to guide clinical decision-making.
Explore This IssueAugust 2020
Background: The poorly characterized and non-specific timeline of symptoms in patients with COVID-19, and the asymptomatic nature of initial infection, may be exacerbating rampant community transmission. The crisis demands a data science-driven approach, with a need to overcome challenges of manual curation of unstructured EHRs and self-reporting outside of the clinical setting.
Study Design: Data study using augmented curation technology.
Setting: Mayo Clinic, Rochester, Minn.
Synopsis: Researchers used deep neural networks over an institution-wide platform for the augmented curation of clinical notes from 77,167 patients subjected to COVID-19 PCR (RNA) testing. They contrasted EHR-derived reports in 26 symptom categories for COVID-19pos versus COVID-19neg patients for the week preceding the testing date. Of these patients, 74,850 tested negative and 2,327 tested positive. Augmented curation technology was leveraged to rapidly curate the COVID-19pos patients’ charts. For charts not yet manually curated, bidirectional encoder representations from transformers (BERT)-based neural networks were used to classify symptoms as being present or not present in each patient. Of their designated symptoms, the researchers identified, in descending order, anosmia/dysgeusia, fever/chills, respiration difficulty, cough, myalgia/arthralgia, and diarrhea as significantly amplified in COVID-19pos over COVID-19neg patients. The study demonstrates how such symptoms can be extracted from highly unstructured institutional knowledge and synthesized using deep learning neural networks. Caveats include the limitations of relying solely on EHR inference to track symptoms and limitations related to data availability.
Citation: Wagner T, Shweta FNU, Murugadoss K, et al. Augmented curation of clinical notes from a massive EHR system reveals symptoms of impending COVID-19 diagnosis. eLife. 2020;9:e58227.