An evaluation of the performance of a NoSQL document database in a simulation of a large scale Electronic Health Record (EHR) system

Ercan, Mehmet Zahid (2017) An evaluation of the performance of a NoSQL document database in a simulation of a large scale Electronic Health Record (EHR) system. [Thesis (PhD/Research)]

[img]
Preview
Text (Whole Thesis)
Ercan_2017_whole.pdf

Download (3922Kb) | Preview

Abstract

Electronic Healthcare Record (EHR) systems can provide significant benefits by improving the effectiveness of healthcare systems. Research and industry projects focusing on storing healthcare information in NoSQL databases has been triggered by practical experience demonstrating that a relational database approach to managing healthcare records has become a bottleneck. Previous studies show that NoSQL databases based on consistency, availability and partition tolerance (CAP) theorem have significant advantages over relational databases such as easy and automatic scaling, better performance and high availability. However, there is limited empirical research that has evaluated the suitability of NoSQL databases for managing EHRs. This research addressed this identified research problem and gap in the literature by investigating the following general research: How can a simulation of a large EHR system be developed so that the performance of NoSQL document databases comparative to relational databases can be evaluated?

Using a Design Science approach informed by a pragmatic worldview, a number of IT artefacts were developed to enable an evaluation of performance of a NoSQL document oriented database comparative to a relational database in a simulation of a large scale EHR system. These were healthcare data models (NoSQL document database, relational database) for the Australian Healthcare context, a random healthcare data generator and a prototype EHR system. The performance of a NoSQL document database (Couchbase) was evaluated comparative to a relational database (MySQL) in terms database operations (insert, update, delete of EHRs), scalability, EHR sharing and data analysis (complex querying) capabilities in a simulation of a large scale EHR system, constructed in the cloud environment of Amazon Web Services (AWS). Test scenarios consisted of a number of different configurations ranging from 1, 2, 4, 8 and 16 nodes for 1Million, 10 Million, 100 Million and 500 Million records to simulate database operations in a large scale and distributed EHR system environment.

The Couchbase NoSQL document database was found to perform significantly better than the MySQL relational database in most of the test cases in terms of database operations -insert, update, delete of EHRs, scalability and EHR sharing. However, the MySQL relational database was found to perform significantly better than the Couchbase NoSQL document database for the complex query test that demonstrates basic analysis capabilities. Furthermore, the Couchbase NoSQL document database used significantly more disk space than the MySQL relational database to store the same number of EHRs.

This research made a number of important contributions to knowledge, theory and practice. The main theoretical contribution to design theory was the design and evaluation of a prototype EHR system for simulating database management operations in a large scale EHR system environment. The prototype EHR system was underpinned by the development of two data models with data structures designed for a NoSQL document database and a relational database and a random healthcare data generator which were based on Australian Healthcare data characteristics and statistics. The design of a data model for EHRs for a NoSQL document database using an aggregated document modelling approach provided an important contribution to data modelling theory for NoSQL document databases using de-normalisation and document aggregation. The design of a random healthcare data generator was another important contribution to design theory and was based on a data distribution algorithm (multinomial distribution and probability theory) informed by National Health Data
Dictionary and published Australian Healthcare statistics. The prototype EHR system allowed this study to demonstrate through a simulated performance evaluation that a NoSQL document database has significant and proven performance advantages over relational databases in most of the database management test cases. Hence this study demonstrated the utility and efficacy of a NoSQL document database in the simulation of a large scale EHR system. This research has made a number of important contributions to practice. Foremost is that the IT artefacts (namely, a data model for storing EHRs in a NoSQL document database, a random healthcare data generator and a prototype EHR system) developed and evaluated in this research can be readily adopted by practitioners. Another important practical contribution of this research is that it is based on the open source availability of NoSQL database and relational database alternatives. Hence, this research can provide a sound basis for lower-income countries as well higher-income countries to establish their own cost-effective national EHR systems without the restrictions, limitations, complexity or complications of similar proprietary relational database systems.


Statistics for USQ ePrint 34469
Statistics for this ePrint Item
Item Type: Thesis (PhD/Research)
Item Status: Live Archive
Additional Information: Doctor of Philosophy (PhD) thesis.
Faculty/School / Institute/Centre: Current - Faculty of Business, Education, Law and Arts - School of Management and Enterprise
Supervisors: Lane, Michael; Gururajan, Raj
Date Deposited: 10 Jul 2018 04:24
Last Modified: 29 Jul 2019 00:00
Uncontrolled Keywords: NoSQL databases, Electronic Health Record (EHR), healthcare systems, relational databases, distributed systems, ACID, CAP theorem, BASE
Fields of Research : 08 Information and Computing Sciences > 0805 Distributed Computing > 080599 Distributed Computing not elsewhere classified
08 Information and Computing Sciences > 0806 Information Systems > 080604 Database Management
11 Medical and Health Sciences > 1117 Public Health and Health Services > 111711 Health Information Systems (incl. Surveillance)
Identification Number or DOI: doi:10.26192/5c09b7aff0cc4
URI: http://eprints.usq.edu.au/id/eprint/34469

Actions (login required)

View Item Archive Repository Staff Only