Abstracting social determinants of health and COVID risk factors from clinical text using NLP

For UCSF Faculty and Staff: Please login via MyAccess to request access to update your Project.

Public Summary

Information about patient risk factors for COVID-19 infection and disease severity are primarily captured by electronic health records systems in the form of clinical notes. To answer important questions about the epidemiology of COVID-19, we need computational tools to 'mine' this data out of the free-text; that is, to process free-text into well-organized data for ultimate use in a variety of research studies. We propose to use a variety of AI tools in order to perform this task in an accurate way that does not otherwise require the efforts of many human annotators.