Developing and Validating Algorithms a Natural Language Processing Algorithm for Caregiver Availability Among Patients with Alzheimer’s Disease and Related Dementia

​Patients with Alzheimer’s disease and related dementia (ADRD) are at higher risk of adverse health events, including 30-day hospital readmission,1 emergency department visits,2 and mortality.3 Caregiver support has been identified as a salient component of care for ADRD patients;4 therefore, there is a critical need for healthcare systems to identify patients lacking support. These critical data elements for ADRD patients are not collected in structured Electronic Medical Records (EMRs), and primary data collection via questionnaires is time and labor-intensive, distracting from the focused care ADRD patients require. Unstructured EMRs (i.e. physician and operative notes) can be leveraged to obtain this information. We will use Natural Language Processing (NLP) that has been successfully implemented to extract value-added information in other settings and populations.5 We will use NLP to extract data from physician and operative notes to collect information on caregiver support relevant for ADRD patients.