Abstract: Epidemiological studies have used a variety of methods to determine biologically implausible values (BIVs) in height, weight, and body mass index measurements. As the number of EHR records available for secondary use has grown, there is significant interest in applying these BIV detection methods to extremely large EHR datasets. We timed and analyzed the performance of several BIV detection methods on the Explorys master dataset, as conducted on a 243 node MR cluster.
Learning Objective 1: Attendees will understand the various factors that impact the implementation feasibility of implausible value detection algorithms and be able to compare several published methodologies.
John Borsi (Presenter)
IBM Watson Health
Wei Yao, IBM Watson Health
Nnenna Ibeanusi, IBM Watson Health