An important trend in modern medicine is towards individualisation of healthcare to tailor care to the needs of the individual. This makes it possible, for example, to personalise diagnosis and treatment to improve outcome. However, the benefits of this can only be fully realised if healthcare and ICT resources are exploited (e.g. to provide access to relevant data, analysis algorithms, knowledge and expertise). Potentially, grid can play an important role in this by allowing sharing of resources and expertise to improve the quality of care. The integration of grid and the new concept of bioprofile represents a new topic in the healthgrid for individualisation of healthcare. A bioprofile represents a personal dynamic "fingerprint" that fuses together a person's current and past bio-history, biopatterns and prognosis. It combines not just data, but also analysis and predictions of future or likely susceptibility to disease, such as brain diseases and cancer. The creation and use of bioprofile require the support of a number of healthcare and ICT technologies and techniques, such as medical imaging and electrophysiology and related facilities, analysis tools, data storage and computation clusters. The need to share clinical data, storage and computation resources between different bioprofile centres creates not only local problems, but also global problems. Existing ICT technologies are inappropriate for bioprofiling because of the difficulties in the use and management of heterogeneous IT resources at different bioprofile centres. Grid as an emerging resource sharing concept fulfils the needs of bioprofile in several aspects, including discovery, access, monitoring and allocation of distributed bioprofile databases, computation resoiuces, bioprofile knowledge bases, etc. However, the challenge of how to integrate the grid and bioprofile technologies together in order to offer an advanced distributed bioprofile environment to support individualized healthcare remains. The aim of this project is to develop a framework for one of the key meta-level bioprofile applications: bioprofile analysis over grid to support individualised healthcare. Bioprofile analysis is a critical part of bioprofiling (i.e. the creation, use and update of bioprofiles). Analysis makes it possible, for example, to extract markers from data for diagnosis and to assess individual's health status. The framework provides a basis for a "grid-based" solution to the challenge of "distributed bioprofile analysis" in bioprofiling. The main contributions of the thesis are fourfold: A. An architecture for bioprofile analysis over grid. The design of a suitable aichitecture is fundamental to the development of any ICT systems. The architecture creates a meaiis for categorisation, determination and organisation of core grid components to support the development and use of grid for bioprofile analysis; B. A service model for bioprofile analysis over grid. The service model proposes a service design principle, a service architecture for bioprofile analysis over grid, and a distributed EEG analysis service model. The service design principle addresses the main service design considerations behind the service model, in the aspects of usability, flexibility, extensibility, reusability, etc. The service architecture identifies the main categories of services and outlines an approach in organising services to realise certain functionalities required by distributed bioprofile analysis applications. The EEG analysis service model demonstrates the utilisation and development of services to enable bioprofile analysis over grid; C. Two grid test-beds and a practical implementation of EEG analysis over grid. The two grid test-beds: the BIOPATTERN grid and PlymGRID are built based on existing grid middleware tools. They provide essential experimental platforms for research in bioprofiling over grid. The work here demonstrates how resources, grid middleware and services can be utilised, organised and implemented to support distributed EEG analysis for early detection of dementia. The distributed Electroencephalography (EEG) analysis environment can be used to support a variety of research activities in EEG analysis; D. A scheme for organising multiple (heterogeneous) descriptions of individual grid entities for knowledge representation of grid. The scheme solves the compatibility and adaptability problems in managing heterogeneous descriptions (i.e. descriptions using different languages and schemas/ontologies) for collaborated representation of a grid environment in different scales. It underpins the concept of bioprofile analysis over grid in the aspect of knowledge-based global coordination between components of bioprofile analysis over grid.

Document Type


Publication Date