Abstract: Eldercare monitoring using non-wearable sensors is an emerging solution for improving care and reducing costs. However, attaching clinical significance to the large amount of data generated by the sensor network is challenging for the clinical personnel. Using electronic health records (EHR) in conjunction with a monitoring system enables sensor data annotation with clinical terms that could ease clinicians’ interpretation. The question is if a nursing home EHR contains enough clinical information to enable such a task? We performed a semantic analysis of a dataset that consisted of clinical records of 54 residents from Tiger Place, an aging in place facility, collected over five years. We found an average of 1.5 clinical terms/resident/day available for sensor data annotation.

Learning Objective 1: Using natural language processing in medical records analysis


Zhongkai Huang, University of Missouri
Mihail Popescu (Presenter)
University of Missouri

Presentation Materials: