NUHS to build supercomputing infrastructure for hospital AI
Credit: NUHS
The National University Health System, one of the three national health clusters in Singapore, has entered into a collaborative agreement with the government-backed National Supercomputing Centre Singapore to build a supercomputing infrastructure to support artificial intelligence programmes in public healthcare institutions by mid-2022.
Meanwhile, it has also partnered with telecommunications giant Singtel to deploy a 5G indoor network with multi-access edge compute capabilities in the operating theatres and wards of the National University Hospital.
WHY IT MATTERS
Hospitals produce large volumes of data each year and modern technologies, such as AI, machine learning, and automation, help medical researchers make sense of these huge amounts of data to improve patient outcomes. Supercomputers can exponentially enhance these tools to build more complex models that can accommodate big data, the NUHS said.
According to a press statement, the supercomputing infrastructure called PRESCIENCE will train AI models that predict patient health trajectories and tell when a patient's condition may deteriorate.
Dr Ngiam Kee Yuan, chief technology officer at NUHS, mentioned that it usually takes days to train AI models with big data "but the new supercomputer could help to cut our training times down to hours allowing our medical and para-medical staff to optimise patient trajectories and to improve the quality of patient care".
Meanwhile, the 5G network project with Singtel would address present limitations in latencies and bandwidth. High-speed connectivity via 5G will also enable better healthcare experiences, such as "smoother teleconsultations, augmented surgical navigation using mixed reality devices, and robot AI capabilities using cloud and edge computing," the NUHS noted.
THE LARGER TREND
NUHS's Dr Ngiam has said that AI is at the centre of the hospital group's digital transformation. An example is its latest AI production platform called Endeavour AI that streams data and processes these data to produce outputs in real-time. It runs with NVIDIA DGX A100 at its core, which enables its AI tools to run quickly in the background and absorb data on a daily basis. Four applications from TIBCO Software were also employed in the AI platform to support the integration of medical data from EMR systems in real-time.