An Automatic Data Mapping for Interoperability of OpenEMR Medical Practice Management Software Using the Fast Healthcare Interoperability Resources

Abstract

Data compatibility in Electronic Medical Records (EMR) among healthcare facilities is necessary, especially for medical practitioners such as doctors or physicians, so that they can grant a more accurate decision on what treatments should be carried out for their patients, since a precise treatment or medication will increase the chance that patients would successfully heal from their disease. The compatibility of EMR data can also be called interoperability. This research attempts to apply interoperability of healthcare data by implementing an automatic mapper of an EMR data from one EMR management system called OpenEMR so that its data can meet the FHIR (Fast Healthcare Interoperability Resources) standard. Specifically, a classifier to categorize the OpenEMR data into the appropriate FHIR format is discussed in this paper. There are three classifiers developed in Java and Python, which utilize the concepts of machine learning classification techniques; in this case, Naïve-Bayes and Decision Tree. Implementations of both machine learning algorithms showed a classification accuracy of 100%, which resulted in the additional implementation of rule-based technique, which also resulted in 100% accuracy. After running similar tests on all three implementations, the results infer that the rule-based technique is better than Naïve-Bayes for development in Java programming language.