The project hopes to demonstrate that public cloud resources can be harnessed to handle large-scale, high-fidelity simulations for medical applications.
Harvard University has announced a medical research program with joint sponsorship from Google Cloud and Citadel Securities. The professors aim to conduct research which would normally be run through a supercomputer on the public cloud, through the use of high performance computing (HPC) services.
Supercomputers are very sought after in the medical industry, because of sophistication for complex calculations. However, there are only a handful of computers in the United States able to run the billions of calculations which are needed to mimic the conditions of medical research accurately.
Aside from the lack of availability of these supercomputers, researchers, especially in academia, often don’t have the resources to pay for access to these supercomputers. This is why Harvard and others are now turning to public cloud resources, even though traditionally the public cloud has not been suitable for large-scale research projects.
Creating digital twins using Google Cloud
The Harvard professors aim to leverage Google Cloud to create digital twins of a geometrically accurate representation of complex blood vessels, along with realistic blood flow and structure, and magnetically controlled artificial bacterial flagella. This is a very complex procedure, but if the results are positive, it could be a boon for researchers looking at the public cloud as a testbed for complex simulations.
See also: AI’s Potential in Enhancing Chronic Disease Management for Improved Patient Outcomes
“Today, the level of computing power required to run simulations of this complexity and at this scale is available from only a handful of supercomputers around the world,” said Petros Koumoutsakos, leader of the research and professor at the Harvard John A. Paulson School of Engineering and Applied Sciences. “With the support of Citadel Securities and Google Cloud, we aim to demonstrate that public cloud resources can be harnessed to handle large-scale, high-fidelity simulations for medical applications. In doing so, we hope to show that easy access to massively available cloud computing resources can significantly reduce time to solution, improve testing capabilities and reduce research costs for some of humanity’s most pressing problems.”
Virtual simulations are being adapted in more industries than just medical research, with construction, engineering, and cybersecurity all looking at ways to reduce the costs of real-world testing through digital twins. By showing that this can be done on the public cloud, it also frees up the need to acquire lots of independent computing power, opening up the market to more businesses and research teams.
“Google Cloud’s high performance computing technologies solutions are purpose-built to both simplify and scale the largest, most complex workloads, enabling researchers to dramatically accelerate time to discovery and impact,” said Bill Magro, chief technologist of HPC at Google Cloud, in an online statement.
Initial tests conducted by the Harvard research team found that cloud resources can achieve 80 percent of the efficiency of dedicated supercomputers. The team will continue to run types of code through the public cloud to see which parts of the research are more conducive to public cloud operations, and which parts may have to continue to rely on supercomputers to perform at a high level.
