01 High-speed networks for e-Health applications Andrew Simpson, David Gavaghan, Lee Momtahan, David Power, Mark Slaymaker 02 Outline • The field of e-Health is a key area in which e-Science and Grid computing technologies can be deployed in anger—and deliver genuine benefits • Essential non-functional concerns, such as security, the expedient delivery of data and quality of service, make e-Health applications appropriate candidates for validating and demonstrating the potential of hybrid optical networks 03 Outline • We have been tasked with demonstrating the benefits of UK Light in the e-Health arena, and plan to adopt a two-pronged approach • The justification for this is that Grid-based e-Health applications are some distance from being deployed ‘for real’ within the NHS: there are many technical, social, legal and ethical issues that must first be overcome • Thus, one aspect of the task will be the consideration and documentation of these issues, while the other will be the utilisation of UK Light to provide tangible benefits to an e-Health project that is currently ongoing: the Integrative Biology project 04 The generic: issues and use cases • In considering the generic issues pertaining to the utilisation of high-speed networks within e-Health applications, we will engage with the e-Health community to capture potential use cases • As a motivating example, consider the possibility of remote breast screening, which is drawn from the experiences of the e-DiaMoND project 05 An example application: remote breast screening • In the United Kingdom, the Breast Screening Programme (BSP) currently invites women between the ages of 50 and 64 to attend a screening session every three years, with subsequent recalls to an assessment clinic if necessary; approximately 1.5 million women are screened by the BSP each year • It is intended that the programme will be extended to include women up to and including the age of 70 by 2004 • This is expected to lead to an increase in numbers to 2.5 million women per year by 2005 • Given that by the end of 2005 every woman screened will have two views per breast taken, this will result in approximately 10 million mammograms per year being taken by the BSP 06 An example application: remote breast screening • It is currently the case that workloads from some clinics are physically transferred to remote clinics for reading • In addition, the BSP is exploring the possibility of compensating for a shortfall in radiologists in the United Kingdom by utilising appropriate expertise from overseas 07 An example application: remote breast screening • A typical screening centre will screen approximately 100 women per day: the average woman will be having her fourth scan, so will have three previous screening sessions’ worth of data • If a clinic were to subcontract screening work, we can expect that it would do so on a batch basis, i.e., a day’s work, which is 100 patients’ worth of data: on average, this would mean three previous sets of four images plus one set of four images, i.e., 16 images, for each patient • In general, a radiologist will perform in the region of 100 readings per one-hour session, which amounts to approximately 100GB of data per reading session • Many of the ethical, security and technical issues pertaining to this scenario will be generic 08 The generic: ethical issues • In the UK, raw data, i.e, identifiable patient data can be moved between clinics if and only if that data is being moved for someone to review it in the interests of ‘direct care’, i.e., it is aiding in the treatment of a patient • Usual data protection rules apply in that the receiver of the data should only keep what they have been provided for a limited time (suitable to perform the task being asked to do) and clearly with a system this would be managed by giving the clinician temporary rights to do something 09 The specific: Integrative Biology • The Integrative Biology (IB) project involves the use of state-of-the-art HPC facilities to enable simulation of very complex biological systems • For example, certain whole-heart models can involve finite element meshes with tens of millions of nodes, and up 50 solution variables • Solution in real-time for just a single heartbeat requires (at least) millisecond temporal accuracy, resulting in data generation in the 10–100 Terabyte range from a single simulation 10 The specific: Integrative Biology To make optimal use of HPC facilities the IB project is developing and extending toolkits to support: • Computational steering • Collaborative visualisation • Performance control • Data curation • Data mining One of the IB project goals is to investigate the potential of using heterogeneous HPC resources within a single simulation 11 The specific: Integrative Biology • In modelling heart mechanics, the equations that model the mechanical pumping action are coupled to the electrical activity generating the muscular contractions via a large system of ordinary differential equations (up to 50) which model the ion flows through the cell membrane • A potentially attractive approach is to solve the ODE systems on an MPI HPC architecture, whilst simultaneously solving the mechanical problem on an OpenMP architecture: to make this approach feasible would require very rapid communications, and the facility to co-schedule the simulations, between the HPC resources 12 The specific: Integrative Biology With respect to the possibility of allowing models to be run at more than one site, even when they are tightly coupled, the extent to which the benefits outweigh the overheads are open to question (it requires co-scheduling of resource in different sites): the answering of this open question will represent a tangible benefit for the IB project 13 Tasks • The collection of use cases and the undertaking of appropriate analyses to determine the legal, ethical, and technical constraints involved with the transfer of medical information • The capture and formal description of non-functional requirements for the transfer of data within IB • The development of application-level monitoring software to monitor adherence to these non-functional requirements within IB • The utilisation of some of the IB tools over UK Light to conduct performance control and collaborative steering • The documentation of lessons learned from the Integrative Biology experience to feed back to the wider community 14 Wider engagement • With respect to the generic, we will be seeking engagement with, and use cases from, the wider e-Health community • The recently announced MRC-funded e-Science projects are particularly relevant in this respect • In addition, we will take advantage of the proposed DTI-funded KTN for Inter-Enterprise Computing to investigate further links • With respect to the specific, we will attempt to link to the HPC group within the ESLEA project, as they are likely to be looking at similar issues: mutual benefits will accrue as our group benefits from their expertise and their group benefits from an additional realistic application for their work 15 Summary • Outline • The generic: issues and use cases • An example application: remote breast screening • The generic: ethical issues • The specific: Integrative Biology • Tasks • Wider engagement