Abstract
Clinical research is becoming ever more collaborative with multi-centre trials now a common practice. With this in mind, never has it been more important to have secure access to data and, in so doing, tackle the challenges of inter-organisational data access and usage. This is especially the case for research conducted within the brain injury domain due to the complicated multi-trauma nature of the disease with its associated complex collation of time-series data of varying resolution and quality. It is now widely accepted that advances in treatment within this group of patients will only be delivered if the technical infrastructures underpinning the collection and validation of multi-centre research data for clinical trials is improved. In recognition of this need, IT-based multi-centre e-Infrastructures such as the Brain Monitoring with Information Technology group (BrainIT - www.brainit.org) and Cooperative Study on Brain Injury Depolarisations (COSBID - www.cosbid.de) have been formed. A serious impediment to the effective implementation of these networks is access to the know-how and experience needed to install, deploy and manage security-oriented middleware systems that provide secure access to distributed hospital based datasets and especially the linkage of these data sets across sites. The recently funded EU framework VII ICT project Advanced Arterial Hypotension Adverse Event prediction through a Novel Bayesian Neural Network (AVERT-IT) is focused upon tackling these challenges. This chapter describes the problems inherent to data collection within the brain injury medical domain, the current IT-based solutions designed to address these problems and how they perform in practice. The authors outline how the authors have collaborated towards developing Grid solutions to address the major technical issues. Towards this end we describe a prototype solution which ultimately formed the basis for the AVERT-IT project. They describe the design of the underlying Grid infrastructure for AVERT-IT and how it will be used to produce novel approaches to data collection, data validation and clinical trial design.
Top1. Introduction
Traumatic brain injury (TBI), also known as head injury, is a significant clinical problem. The incidence of severe TBI is approximately 200 patients/100,000 population with the most common causes including road traffic accidents, falls and assaults. Males are more than twice as likely to receive a severe injury than woman and currently the reported mortality rate following severe TBI ranges from less than 10% up to 50% with the most common rate quoted between 20-30%. Although the incidence of TBI is significantly less than those of other major medical diseases such as cardiovascular disease, cancer and stroke, as TBI occurs mostly in the young and the resultant morbidity is severe and long-lasting, the burden of TBI to the individual, their carers and the society that supports them is as great if not greater than the other disease domains. On a European level, fifty percent of the years individuals spend with disability are caused by brain disease of which traumatic brain injury now carries an equal burden to patients as do those of cerebrovascular and depressive illness disorders (Olesen 2003).
After ten years of pharmaceutical industry sponsored drug development and despite promising pre-clinical data, most of the clinical trials of these agents have failed to show any significant improvement in patient outcome (Narayan 2002). Many researchers feel a significant cause underlying this lack of success is the poor resolution of paper based methods for detection of adverse events and poor methods for monitoring of and controlling for protocol violations and medication errors. These limitations combine to make it difficult to detect small but clinically important treatment effects in the general noise of the brain injured patient management environment. The poor success rate of TBI clinical trials combined with the high cost to the pharmaceutical industry to conduct phase III trials in brain injury has, in recent years, caused a reluctance of the pharmaceutical industry to bring forward promising compounds to clinical trial in the field of brain injury. The high cost of conducting clinical trials is due in large part for the need to hire specially trained staff to collect and validate data. If technical solutions could be developed to reduce, even partially, the need for human resources in the data collection/validation process, potentially enormous savings could be made by these organisations. These efficiencies would ensure the organisations’ longer term sustainability, and the lower running costs would reduce the cost of service delivery. Above all, this would improve the overall patient care.
This chapter focuses upon how Grid based infrastructures can help to address these issues. We focus in particular on the aspects of usability and security of Grid based e-Infrastructures and illustrate with examples from a range of projects at the National e-Science Centre at the University of Glasgow, how the vision of the Grid in providing seamless access to a range of heterogeneous resources (such as a variety of neurological data resources) can be undertaken in a secure, ethical framework where information governance and associated policy is paramount.
Key Terms in this Chapter
Authorisation: The process of restricting access to resources only to those permitted to use them. In the security domain this will typically entail the definition of security policies associated with resources, the assignment of privileges to individuals that should be able to access those resources and the subsequent enactment of those policies when requests for access are received by individuals. Standards have been identified to support both the definition of policies, where access decisions need to be enforced, and how such decisions are made. Authorisation typically augments authentication and allows finer grained access control to be supported.
Hypotension: Abnormally low blood pressure which can be especially dangerous for brain injury patients.
Single Sign-On: The ability to securely access and use a variety of distributed resources without the need for multiple usernames/passwords or authentication challenge/responses. In the Grid world this is typically achived through trust of the CA that issued the certificate and local policy on whether that individual with that certificate is allowed access. For many Grids, this can be through a mapping of the Distinguished Name associated with the certificate to a local system account.
Registration Authority (RA): Typically an individual at an institution that supports the processes required by a CA to verify the identity of individuals applying for digital certificates. Typically this is through the certificate requestor presenting in person a visual identity such as a passport or student matriculation card to the RA.
Glasgow Coma Score (GCS): Widely used clinical score assessing brain stem function following a suspected brain injury.
Authentication: The act of establishing or confirming something or someone as authentic. In the security domain this might for example involve electronically confirming the identity of a person wishing to access a given software or hardware resource. Authentication can be achieved in many ways, e.g. usernames/passwords, certificate based systems etc.
Pressor: Pharmacological agent which acts to increase the work of the heart and often includes actions on increasing systemic vascular resistance towards raising blood pressure when hypotensive.
Certificate Authority (CA): An entity which issues digital certificates for use by other parties (including individuals and computers). Through trusting a CA and the process by which it issues and revokes certificates, the certificates can be used for accessing multiple resources seamlessly to support one of the basic tenets of Grid: single sign-on.
Adverse Hypotensive Event: Hypotension lasting 5 minutes or longer falling below the commonly accepted threshold of 90 mmHg systolic OR 70 mmHg mean pressure.