Use this URL to cite or link to this record in EThOS:
Title: An experimental study and evaluation of a new architecture for clinical decision support : integrating the openEHR specifications for the Electronic Health Record with Bayesian networks
Author: Arikan, S. S.
ISNI:       0000 0004 8497 5357
Awarding Body: UCL (University College London)
Current Institution: University College London (University of London)
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Healthcare informatics still lacks wide-scale adoption of intelligent decision support methods, despite continuous increases in computing power and methodological advances in scalable computation and machine learning, over recent decades. The potential has long been recognised, as evidenced in the literature of the domain, which is extensively reviewed. The thesis identifies and explores key barriers to adoption of clinical decision support, through computational experiments encompassing a number of technical platforms. Building on previous research, it implements and tests a novel platform architecture capable of processing and reasoning with clinical data. The key components of this platform are the now widely implemented openEHR electronic health record specifications and Bayesian Belief Networks. Substantial software implementations are used to explore the integration of these components, guided and supplemented by input from clinician experts and using clinical data models derived in hospital settings at Moorfields Eye Hospital. Data quality and quantity issues are highlighted. Insights thus gained are used to design and build a novel graph-based representation and processing model for the clinical data, based on the openEHR specifications. The approach can be implemented using diverse modern database and platform technologies. Computational experiments with the platform, using data from two clinical domains - a preliminary study with published thyroid metabolism data and a substantial study of cataract surgery - explore fundamental barriers that must be overcome in intelligent healthcare systems developments for clinical settings. These have often been neglected, or misunderstood as implementation procedures of secondary importance. The results confirm that the methods developed have the potential to overcome a number of these barriers. The findings lead to proposals for improvements to the openEHR specifications, in the context of machine learning applications, and in particular for integrating them with Bayesian Networks. The thesis concludes with a roadmap for future research, building on progress and findings to date.
Supervisor: Ingram, D. Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available