FHIR: Cell-Level Security and Real Time Access with Accumulo

Abstract

The American Recovery and Reinvestment Act (ARRA) requires the adoption of Electronic Medical Records (EMRs) for seventy percent of the primary care provider population by 2014. Furthermore, by 2015 providers are expected to be utilizing EHR in compliance with "meaningful use"[28] definition or they can face financial penalties under Medicare. In addition to this momentous task, EMR data has stringent security requirements. It is largely due to these security requirements that medical information is being digitized. However, sharing to entitled information is often slow or non-existent because of information silos. Fast Healthcare Interoperability Resources (FHIR) is an emerging information sharing standard that is designed to aid in the tearing down of these silos. The focus of this thesis is to show how FHIR can be further improved by allowing for cellular level security. Additionally, this thesis will introduce novel ways that vast amounts of FHIR resources can be stored and queried in real-time with Accumulo. It will do so by utilizing and improving on Dynamic Distributed Dimensional Data Model (D4M) [9] Schema to better allow for "real-time" REST queries of FHIR compliant data. Pagination is necessary for it to remain a real-time system since some queries can have millions or even billions of positive hits. To satisfy this requirement a new approach to Accumulo pagination is laid out that increases performance, flexibility and control. All tests are performed against a M4.2xlarge Amazon Machine Image.