A Framework for Automatic Clustering of EHR Messages Using a Spatial Clustering Approach

Area of Science:

• Health Informatics
• Computer Science

Background:

• Existing Health Level Seven (HL7) standards (v2, v3, CDA) face semantic interoperability challenges and lack support for smart devices.
• Many healthcare organizations use proprietary Electronic Health Record (EHR) formats, hindering migration to modern standards like HL7 Fast Health Interoperability Resources (FHIR).

Purpose of the Study:

• To propose a framework for converting proprietary EHR messages to HL7 v2 format.
• To apply an unsupervised clustering algorithm (DBSCAN) for automatic grouping of HL7 v2 messages.
• To establish a foundation for a generic mapping model to convert data into the HL7 FHIR standard.

Main Methods:

• Developed a framework to convert proprietary EHR messages into HL7 v2 format.
• Utilized the DBSCAN (density-based spatial clustering of applications with noise) algorithm for unsupervised clustering of HL7 v2 messages.
• Experimental validation of the framework's clustering and insight generation capabilities.

Main Results:

• Successfully converted proprietary EHR messages to HL7 v2 format.
• Demonstrated the ability of DBSCAN to automatically group diverse HL7 v2 messages irrespective of their semantic origins.
• Provided analytical insights into the clustered message data.

Conclusions:

• The proposed framework facilitates the conversion of legacy EHR data to HL7 v2.
• Unsupervised clustering aids in organizing and understanding varied healthcare message formats.
• This work supports the development of a generic mapping model for HL7 FHIR, enhancing healthcare data interoperability and leveraging modern technologies.