الفهرس 
Table of ContentOnly 14 pages are availabe for public view
 |  | from | 97 |  |  |
| | | from | 97 | | |
Abstract
With the emergence of modern applications frameworks which need a shorter turnaround time, borders between operational and analytical workload have been blurred. This architecture supports hybrid workload processing within a single logical database instead of two different systems. Electronic health records (EHRs) system which is a quintessential part of healthcare information system (HIS) is progressively being affected by the concept mixed workload. These systems work beyond just recording patients` health data. They have multiple secondary functionalities, such as data reporting and clinical decision support. Their workload usually encompasses two main practices: clinical use, which is regarded as a transactional workload, and research use which is dedicated to analytical workload. As each of these workloads has a different access pattern. They have seemingly contradictory solutions. There is also the operational business intelligence (BI) workload, including ad hoc queries, which is in the middle between primary and secondary uses. This data access pattern tends to be unpredictable; it typically involves reading an extensive amount of data at one time and can include various complex joins. Earlier healthcare data persistence systems were built depending on the relational schema approach. With the rise of the ”no one size fits all” concept, several alternative NoSQL stores have been developed to address the shortcomings of the relational database systems as they are more suitable for meeting the specifications of distributed EHRs systems. Thus, various data modeling approaches have been introduced for medical data persistence according to use case scenarios. These works investigate not only the type of NoSQL store that has to be chosen, but which NoSQL products in that type will be used. This thesis first distinguishes different EHR functions in modern healthcare applications according to their corresponding workload specifications. As both database selection and its related schema architecture are aspects that affect the effective management of healthcare data, particularly in real-time usage systems. Insights about the suitability of the various NoSQL data models as EHRs storage persistence is also made according to this review. The related works’ analysis showed that each database model type is designed for a particular purpose and offers different solutions and performance according to the application context. Flexible document-based NoSQL databases are better than the relational and the native XML storage approaches for managing large-scale EHRs data. They reveal their ability to handle both transactional and analytical workload. The main purpose of this thesis is to propose a unified EHRs data framework that can simplify HIS infrastructure. It investigates the suitability of the document-based NoSQL persistence mechanism, storing EHRs data as a design choice for managing BI workload. The performance of the most popular two document-based NoSQL back-ends, Couchbase Server and MongoDB, is compared for handling varied complexity ad hoc queries according to the size of the database and query execution time. Results showed that while MongoDB can execute simple single-document queries nearly in milliseconds. It does not provide satisfactory response time for unplanned complex queries spanning multiple documents. By utilizing its analytics services and multi-dimensional scaling (MDS) architecture, the response time of multi-node Couchbase cluster outperforms the response time of MongoDB for both simple and complex healthcare data access patterns. The main advantageous of the proposed flexible Couchbase-based healthcare architectural framework is to offer a single, integrated platform that can be used for almost all varied complexity operational workloads as well as operational analytics. It provides different levels of services that could be efficiently adapted according to varied application workload requirements.