����ֱ��

The University of Toronto is teaming up with processor giant AMD to launch a supercomputing platform that will power the university’s health research – including on global threats such as COVID-19.

The initiative, dubbed SciNet4Health, will allow researchers and clinician scientists at ����ֱ�� and its partner hospitals to access and analyze massive databases of patient health information – in a secure way that protects patients’ privacy – using technologies such as machine learning.

SciNet4Health is made possible by , capable of a quadrillion calculations per second. It promises to lead to advancements in vaccine development, drug discovery, genomics research and mathematical modelling.

“The new resources that we are receiving from AMD are going to allow us to set up the computing infrastructure that our health researchers need, especially right now during the time of COVID-19 when many of our faculty are working towards various solutions and positive outcomes for the pandemic,” said Alex Mihailidis, ����ֱ��’s associate vice-president, international partnerships.

“Until today, ����ֱ�� did not have a dedicated computing infrastructure for health researchers that can support patient data, so this is going to have a significant impact on our research.”

SciNet4Health will operate out of the facilities of SciNet, the ����ֱ��-based supercomputing consortium and home to Canada’s most powerful research supercomputer: Niagara. The program will allow SciNet, which has enabled advancements fields ranging from astrophysics to climate science, bring its capacity for cutting-edge data science to health research.

Daniel Gruner, chief technology officer at SciNet, said high-performance computing allows for complex calculations that regular computers simply can’t manage.

“If you’re thinking of using AI and machine learning to try and make sense of huge and diverse data, you need these big computers because it can’t be done on a small machine – it requires a lot of math, a lot of computation, so you need computers that are specially geared towards that,” he said.

“The resources we’re receiving from AMD happen to be very heavy on GPUs [graphic processing units] that can run deep learning calculations a lot faster than a regular CPU can.”

The donation by AMD, based in Santa Clara, Calif., consists of 20 “compute nodes” – individual computers that comprise a high-performance computing cluster – each with eight GPUs.

“That’s a whole lot of power,” Gruner said.  

It’s also power that will be completely in-house. Until now, ����ֱ�� health researchers in need of supercomputing worked through partner initiatives such as HPC4Health, a high-performance computing network established by UHN and the Hospital for Sick Children. SciNet4Health drew on HPC4Health’s experience using patient data to establish its procedures and protocols. The two organizations plan to work together to meet the needs of the health sciences research community in and around Toronto.

“This is helping catalyse our ability to do more private health information research inside the university,” said Gruner.

For his part, Mihailidis says the machine learning and deep learning capabilities that will be provided by SciNet4Health will enable researchers to work with patient data to a degree that wasn’t previously possible due to security and privacy considerations. A professor in the department of occupational science and occupational therapy in the Faculty of Medicine, Mihailidis cited his research on aging and geriatrics as just one example of the kind of work that stands to benefit. 

“We’ve been doing a lot of work around collecting data about what older people are doing in their homes and communities, and using machine learning, deep learning and other predictive analytics to determine changes in their health,” he said.

“The problem we’ve had to date is that because we haven’t had secure servers that have allowed us to securely use patient data, we’ve had to scrub the data to the point where the personal attributes are being removed – and because of that, our predictive models on their health aren’t as accurate as they could be if we were able to include the patient health data itself.

“Having this type of resource at the university will allow us to take that type of research to the next level.”

����ֱ�� is among a small group of universities to receive the supercomputing systems from AMD. Others include Stanford University and the University of California, Los Angeles.

“AMD is proud to be working with leading global research institutions to bring the power of high-performance computing technology to the fight against the coronavirus pandemic,” said Mark Papermaster, AMD’s executive vice-president and chief technology officer.

“These donations of AMD EPYC and Radeon Instinct processors will help researchers not only deepen their understanding of COVID-19, but also help improve our ability to respond to future potential threats to global health.”