Infrastructure hardware and software

multicore2 pbentley

The delivery of genuinely personalised medicine on a European scale requires unprecedented levels of computation on massive amounts of heterogeneous biomedical data. ITFoM proposes solutions to facilitate the emergence of new ICT to benefit medicine, and focuses on providing key underlying hardware and software infrastructure to cater both for the immediate demands of other ITFoM platforms and the increasing clinical deployment of ITFoM technology.

ITFoM is expected to utilise three kinds of compute resources for different types of computation. Large integrated simulations will require powerful parallel supercomputers, for which ITFoM consortium members may have access to PRACE research infrastructure (early discussion has begun with PRACE, in order to present this new need for (usage of) massive computing power). For some other types of computation ITFoM will also be able to utilise capacity computing resources such as cloud or grid. These will be accessible via a single portal managing fast exchange of data, with systems based on existing European infrastructure already in production such as EGI.eu. Finally, specialised computing devices dedicated to fast data processing will be located close to the scientific equipment acquiring data to perform primary data processing.

The demands of ITFoM go beyond the capabilities expected from roadmaps for general purpose HPC technology over the next decade. ITFoM addresses this by combining existing large‑scale computing infrastructure (PRACE, etc.) with customisation and optimisation of key components of the technology chain to enable a giant leap forward in overall computational capability for biomedical data processing, modelling and simulation. Through extensive industry involvement, ITFoM will ensure knowledge transfer to industry and advances in computing technology in Europe. 

Underlying compute and storage capabilities that can match ITFoM’s demands

High performance computer systems built around heterogeneous storage and compute technology will be developed. Customisation and optimisation of hardware and/or software for the ITFoM use cases will deliver order of magnitude improvements in capability. In particular, bio‑informed computing has potential to make otherwise impossible computational challenges practical by mimicking the way the human body itself "computes" in real‑time.

Support for deployment of ITFoM workflows to “run anywhere”

Virtualisation layers will support both HPC clusters and cloud or grid environments. This will be key to deployment of ITFoM on a large scale, allowing the workflow to be executed on a mix of dedicated compute appliances, local HPC clusters, private cloud, grid, remote HPC centres (e.g. PRACE) or public cloud environments.

Protection of patient data and compliance with ethical and regulatory requirements

A security service will provide authentication, authorisation and data access protection across ITFoM datasets.

Efficient and scalable data acquisition and model parameterisation

Dedicated computer systems for primary data analysis will be designed to be located close to data acquisition devices and novel algorithms will address the “inverse problem” of using acquired data to parameterise reference models for a specific individual patient.

Ease of use

Intuitive user interfaces and visualisation technologies will make personalised medicine practical in a clinical environment.

 

A series of component prototypes, the deployment of major components for use in other platforms, and the development of two integrated prototypes deployed at 30 months in hospitals for clinical use will be demonstrated. These aim to bring demonstrators of practical, applied solutions to our stakeholders to successfully answer to the challenge of the practical application of personalised medicine in daily clinical use, define future use models and innovative solutions for industrial applications (value chains of the Drug development, Diagnostics, Medical device, sensors, ICT).