One of the most exciting aspects of supercomputing, for me, is when we step in the world of research and take on some of the great challenges of the ages. Over recent years, advances in high performance computing (HPC) technology have shortened the time between hypothesis and insight for researchers. And now we have a new challenge on which to concentrate our brain and computational power—as we try to power the simulation of digital brains.
Hewlett Packard Enterprise helps the EPFL Blue Brain Project (BBP) to advance the understanding of the brain by supplying the supercomputing power they require to digitally reconstruct and simulate the mammalian brain. Over the last few years, HPE has taken on a number of complex and seemingly impossible challenges: we’ve pioneered Memory-Driven Computing and designed computers to assist inter-planetary missions and take people into the farthest reaches of space—as well as reaching back in time and space, in collaboration with the COSMOS group, to study the beginnings of our universe.
However, reaching inside ourselves and studying the brain is perhaps even more challenging than taking on the galaxies.
The human brain is one of the most complex phenomena in the universe, and its digital reconstruction requires next-generation supercomputers and deep collaboration between brain researchers and computer engineers. As our president and CEO, Antonio Neri, said: “Our mission is to create technologies that improve our quality of life, including powering technologies for the healthcare industry to deliver targeted treatments and save lives, HPE is bringing advanced supercomputing and bespoke applications to empower new research that can have transformative benefits for the neuroscientific community and society at large.”
The Blue Brain Project (BBP) aims to build comprehensive digital models of the brain, which will provide the basis for a potentially unlimited range of simulations, each representing an in-silico experiment. These digital experiments will not only require huge computing power, but also a range of very different computing profiles to support models of the brain’s different levels of organization and their interactions, as well as different types of modelling and simulation methodologies.
As soon as Antonio received a request from our Swiss team to work with BBP, he recognized the importance of this as an extension of our global partnerships to model and understand the human body. He asked me to go to Geneva and, only three days after the initial request, I was lucky enough to be sitting with the BBP team co-designing the system with them. It is particularly exciting as the BBP’s challenge to study the mammalian brain dovetails nicely with two existing projects we are working on:
- The Living Heart, a collaboration between HPE and Stanford University to create multi-scale 3D models of the heart to monitor circulation and to virtually test medications in development and ultimately predict drug-induced arrhythmias, even if the patient is on the other side of the world.
- DZNE is studying a population of 30,000 people over 30 years to find answers for brain diseases like Alzheimer’s, leveraging an HPE supercomputer with Memory-Driven Computing properties to improve the lives of the 1B people around the world living with neurological disorders.
As we are tackling with DZNE, populations around the world are aging, and brain diseases such as dementia and Alzheimer’s are becoming more prevalent. In fact, DZNE studies show that fighting dementia currently costs $1 trillion per year. Understanding the brain and coming up with cures and innovations that will help ease the burden on health providers—while providing people with a better quality of life—is becoming more and more important. And brain-related illnesses are not the only problem. By 2025, 1.2 billion people on Earth will be elderly. According to the World Health Organization, by 2020 chronic diseases such as cancer and diabetes will account for almost three quarters of deaths worldwide.
Until now, the viable option for testing medical hypotheses has been by testing on animals or humans. This has risks and can raise ethical questions; however, it is also more expensive, slower and less accurate than if we can create computer models that can simulate human body functions. It may seem impossible to switch entirely to computer testing, but only a few years ago people would never have imagined getting on board an aircraft that had been completely designed and simulated on a computer. Nowadays, the first aircraft that comes off the production line is the final design, not a prototype, because we have advanced our testing capabilities to the point where we are able to learn as much, if not more, in the computer model than we would from traditional processes. I firmly believe that something similar can be achieved through our work with organizations such as BBP and the Living Heart—we may eventually be able to construct digital models that are more effective than any human or animal tests.
Taking it forwards, we can hope to create “digital twins” of organs like the heart, or even of single cells, for individual patients. Simulations can then be run to find out how different people would react to different treatments. At that point, we will have taken the massive step from, generalized, traditional and sometimes even inaccurate research to the provision of truly personalized medical care with models that can be run at low cost and almost in real time to aid diagnosis and treatment plans. This will augment the advances that we have already made in precision medicine, as HPE is helping doctors to stop thinking of the “average patient” and helping them treat the “actual patient.”
Of course, such incredible ambition needs a quite incredible computer. Modelling an individual neuron at BBP today leads to around 20,000 ordinary differential equations. When modelling entire brain regions, this quickly rises to 100 billion equations that have to be solved concurrently. To provide the massive amount of computing power that will be necessary, BBP will be installing an HPE SGI 8600 supercomputer system in Lugano, Switzerland, and will make use of a cluster comprising 372 compute nodes. The HPE SGI 8600 is a sixth-generation system designed to solve the world’s most complex problems in areas ranging from life, earth, and space sciences, to engineering, manufacturing and national security—meaning that it was specially created for applications like BBP in which the power of big data and AI is harnessed to further our human understanding of the world we live in.
Endeavors like our collaboration with BBP bring a lot of hope for the future. There are great advances in areas like medicine thanks to big data, AI and super-computing, but there are also gaps that are slowing enhancement down—such as the need to digitize more patient records and get the data into a format that can be used. Projects such as this one show that we are pushing past the limitations and using data to accomplish feats that once seemed impossible. If we can do that, then the possible will follow in good time. It is something I am very proud to be a part of.