Document Details

Document Type : Thesis 
Document Title :
USING FAST HEALTHCARE INTEROPERABILITY RESOURCES FOR LINKING ARTHRITIS INTERNET REGISTRY TO MOBILE HEALTH APPLICATION
استخدام موارد قابلية التشغيل البيني السريع لربط سجل الانترنت لالْتِهابُ المَفْصِل‎ بتطبيقات الصحة المحمولة
 
Subject : Faculty of Computing and Information Technology 
Document Language : Arabic 
Abstract : This study embarks on a journey with a two-fold aim: test the feasibility of Fast Health Interoperability Resource (FHIR) as a standard approach for data collection and interoperability; and test the precision and accuracy of multi-class classification models of machine learning in predicting Rheumatoid Arthritis and its root causes. Both of these aims are to help the RA community in reaching decision for an integrated and intelligent information system which can be very important in treating the potential root causes of Rheumatoid Arthritis before it fully develops: one of the intriguing problems for the Arthritis community. The experimental aim shall be to link Electronic Medical Records (EMR) and personal health monitors with national databases and then use the data to make predictive clinical decisions regarding the future health of the patients and provide doctors with a probable root cause which may cause RA in the future time period predicted. This is a multi-class classification problem which requires data to be processed and analyzed from multiple sources which is acquired through a standard approach; given that data is to be trained through machine learning algorithms. The premise of the study is to test the feasibility of FHIR data standards and prediction algorithms most appropriate for the case of Rheumatoid Arthritis community: The tests are carried out through experiments conducted on data acquired from different resources. The system can also be tagged as an integral module for intelligent information system for Rheumatoid Arthritis experts and patients. Linking EMRs with national databases will allow us to create a training dataset, while at the same time is important for the clinical decision-making system as it will enable prediction in real time. In order to link health monitoring data from an App to EMRs and national databases the FHIR API is utilized. As defined, FHIR is a standard currently at draft level and its utilization has never been done before for RA. The efficacy of FHIR will be tested in ensuring interoperability in real time. The evaluation criteria will be the time in model training, precision and accuracy of prediction, for data acquired through FHIR, which is analyzed relative to the requirements gathered from Arthritis specialists and the usability requirements defined through surveys. Simultaneously, a comparison of multiclassification models in machine learning is carried out to demonstrate which algorithm works best for the established business case. The results demonstrate that FHIR is an appropriate interoperability standard for linking databases. The machine learning algorithm is implemented through the use of Microsoft Azure (cloud platform) machine learning studio. The results of the models implemented show that the most appropriate machine learning algorithm for the combined data is the Multiclass decision tree algorithm, which shows the accuracy and precisions within the acceptable limits as pointed out by experts and patients. 
Supervisor : Prof. Abdullah Saad AL-Malaise AL-Ghamdi 
Thesis Type : Master Thesis 
Publishing Year : 1440 AH
2018 AD 
Added Date : Monday, November 19, 2018 

Researchers

Researcher Name (Arabic)Researcher Name (English)Researcher TypeDr GradeEmail
منال سعيد ابوملحهAbumelh, Manal SaeedResearcherMaster 

Files

File NameTypeDescription
 43829.pdf pdf 

Back To Researches Page