AI and Machine Learning

Using EHR Metadata to Determine Relative Value of Care

Wednesday, August 11 at 2:45 PM - 3:45 PM PDT
Caesars, Forum 131
Pre-2021 Centers for Medicare & Medicaid Services evaluation and management guidelines for calculating billing levels of service require documentation of specified bullet points for each level of service designed to counter fraud and abuse, and standardize the methodology to more objectively determine the appropriate value of the visit. 2021 guidelines will be time-based or require assessment of medical decision-making. With widely deployed EHRs, there are opportunities for technology to provide a better solution. In this research, we are using machine learning techniques over EHR data and metadata found in audit logs to automatically and accurately determine the appropriate level of service provided, as an alternative to documentation-based calculations or subjective decision-making criteria or time-based calculations that have the potential for fraud. In an effort to compensate appropriately, objectively and consistently for services, this approach can capture criteria to assure payment integrity without increasing burden.

Learning Objectives

  • Describe the challenges of working with EHR metadata
  • Recognize the patterns seen in use of EHRs based on audit log data
  • Compare metadata across EHR vendor products
ABPM 1.0, ACPE 1.0, CAHIMS 1.0, CME 1.0, CNE 1.0, CPHIMS 1.0
Government or Public Policy Professional, Payer, Clinical Informaticists


Sarah Corley, MD, FACP, FHIMSS
Chief Medical Advisor
The MITRE Corporation
Neeraj Koul, PhD
Senior Principal
The MITRE Corporation