Skip to main content
European Commission logo print header

A Centre of Excellence in Computational Biomedicine

Periodic Reporting for period 1 - CompBioMed2 (A Centre of Excellence in Computational Biomedicine)

Période du rapport: 2019-10-01 au 2021-03-31

Computational methods, based on human biology, are now reaching maturity in the biomedical domain, rendering predictive models of health and disease increasingly relevant to clinical practice by providing a personalised aspect to treatment. Computer based modelling and simulation is well established in the physical sciences and engineering, where the use of high performance computing (HPC) is now routine.

The major purpose of our Centre of Excellence is to promote and foster the use of HPC as a fundamental cornerstone of computationally assisted biomedical research and help translate this into medical and clinical practice. We have, therefore, invested substantially in community building to spread knowledge, tools and best practice to students, researchers, and decision makers across the domain and to future and present clinicians. HPC has the potential to enhance industries in the healthcare sector including pharmaceuticals and medical device manufacturers, and underpinning a range of emerging sectors, such as those concerned with e-health and personalised medicine. The innovative modelling and simulation techniques we develop within this Centre are proving to be of great interest and relevance to industrial researchers (including medical device manufacturers), HPC manufacturers and independent software vendors as well as with clinical practitioners.
We have deployed and scaled up several application codes across High Performance Computing (HPC) centres, made possible through compute allocations awarded to CompBioMed. Selected codes from our partners are tested on the largest supercomputers in the world and work with the computer centres as part of co-design projects which are destined to produce well attuned exascale machines in the future (e.g. MEEP). Cardiovascular applications are modelling the full arterial tree and working towards whole-human scale. We have investigated the effect on blood flow when plaques are formed and how the insertion of a stent or conducting a bypass graft can affect a patient’s prognosis. Heart models have been used at various scales from single cell to medical device trials and been key in investigating potential drug candidates for COVID-19. In the Molecularly based Medicine Exemplar, collaborations have investigated drug candidates against COVID-19 and developed new workflows using machine learning and molecular dynamics applications. The Neuro-musculoskeletal Exemplar has used bone modelling software developed within the project in a collaboration with Sheffield Teaching Hospital to investigate how the angle of a fall can affect bone fractures. This has also been extended to look at the risk of fracture over a 10-year period. Our new partner in Bologna (UNIBO) is using a new model to investigate Bone Strength and will conduct in silico trials on over 1000 patients.

Our most high-profile dissemination activity from CompBioMed1 was the production and screening of an IMAX film called “The Virtual Human” (https://youtu.be/1FvRSJ9W734) which we intend to follow up with another film in CompBioMed2. We are repeating our large-scale conference over 3 days in September 2021 in a virtual environment and are well advanced in the planning of this. We have already published over 70 scientific papers and our partners have participated in over 70 times in major conferences and workshops and organised two workshops. Our media and social media activity has been prolific, reaching collective audience sizes in excess of 25 million people (far beyond what we were able to achieve in CompBioMed1). This was owed in part to social media posts about our work from heavily followed twitter accounts, along with appearances on television such as euronews. We have extended our courses for UCL’s medical degree programme and biomedical students to the University of Sheffield (USFD) and have plans to extend this into more universities in the coming years.

We have established a new External Expert Advisory Board (IAB) by increasing and diversifying its initial membership to include clinicians. We have held two meetings with the EEAB where we have been able to discuss our plans and garner their thoughts to enact in the upcoming year.

We have established a set of metrics for monitoring and reporting computational patterns that will be used on the future exascale machines. To that end, we have identified three main patterns that are used within the Centre of Excellence:
1. Monolithic: deployment of a single computational job spread over a substantial fraction of the compute resources or a single supercomputer, including pipelines necessary to manage such large computational jobs.
2. Coupled: Deployment of multiple, communicating subcomponents each assigned to a sub-section of compute resources, in total comprising a substantial fraction of the supercomputer.
3. Ensemble: Multiple instances of an application launched in parallel with different input data; each such instance may be a monolithic parallel code, itself running at extreme scale. This pattern now encompasses highly complex data intensive workflows combining conventional HPC applications with machine learning and/or AI components.

We have assessed the Data and Analysis requirements of the consortium and beyond, especially with respect to running large computational jobs on a machine and being able to access the data produced. This includes communication of this data between centres, storage and analysis. To aid with this, we have established and strengthened collaborations and joint projects with other European initiatives such as LEXIS, DICE (part of the EOSC project), EUDAT and MEEP. This has also spread to our service provisions, which have grown during the first period of CompBioMed2, and our access mechanisms for these services has been optimised. We have established a channel in which people can get access to the top knowledge within the CoE to find out about scaling possibilities for their codes, and how our partners have scaled our own models.
We have run applications on some of the largest supercomputers in the world, coming very close to a full partition of the machine for one application. By doing this we are learning the bottlenecks and issues that will need to be addressed in our own codes and within the supercomputing network itself, to ensure that we can make use of upcoming exascale computers.

We have published over 70 scientific publications in internationally leading journals. We will continue to work on these applications where a combination of capabilities through translational medicine will see the greatest uptake in coordination with our EEAB and Associate Partner base to broaden the scope of users.

In education and training we are focusing on the core principles of biomedical research, and we are reaching out not only to computational scientists but targeting a new generation of clinicians who will soon be working in hospitals. We have integrated a course into the medical training at UCL and USFD, which has also evolved to allow distance or online learning to be used and will include additional institutes throughout Europe in the coming years. This has also enabled us to establish students in the teacher role, showing the need and appetite for this work within the medical student cohort.

We are working on avenues for sustainability of the CoE beyond the lifetime of the project and will report further in future reports.
Long version of the CompBioMed logo