Using the CareMap with Health Incidents Statistics for Generating the Realistic Synthetic Electronic Healthcare Record

The de-personalised Electronic Healthcare Record (EHR) for secondary use has suffered re-identification. The Realistic Synthetic EHR (RS-EHR) is a promising solution safe from the threat of re-identification. This paper addresses the problem of generating the RS-EHR without using the real EHR by exploiting published Clinical Practice Guidelines (CPG) and Health Incidence Statistics (HIS). The CoMSER method takes a constraint-based approach involving: (1) formalising CPGs into the CareMap constraint and the CareMap into the State Transition Machine (STM), (2) incorporating published HIS-based constraints into the STM, and (3) exploiting domain expertise in verifying domain knowledge and creating the re-usable library of clinical notes. A preliminary evaluation of the CoMSER Method produces the RS-EHR that is considered realistic. The main contribution of this work is the approach that uses an HIS-enriched and CPG-based CareMap for generating RS-EHR with neither access to the real EHR nor using de-personalised EHR.