While the hardware still has catching up to do, quantum-based machine learning is already outperforming classical models.
Jeff Lagasse, Associate Editor
CHICAGO – Quantum computing reached a milestone in 2022 when a 400-plus qubit machine was demonstrated at a time when experts were questioning the feasibility of even a 100 qubit system. The question is no longer whether quantum computing will speed up applications in the world of healthcare – it’s now a matter of when.
A qubit (or quantum bit) is the basic unit of information in quantum computing. The number of qubits matters, because the more qubits, the more computing power can grow exponentially. In terms of healthcare, this has emerging possibilities in the realm of machine learning.
The quantum community has discovered problems that can’t be handled with classical machine learning, but are efficiently solvable on quantum computers. That means it’s only a matter of time before the technology has real-world value.
Dr. Frederik Floether, lead quantum and deputy CEO of QuantumBasel, and Numan Laanait, senior director of engineering at Elevance Health, told an audience at the HIMSS23 global conference in Chicago Wednesday that quantum computers are based on a model entirely different than that of its classical counterparts.
“It’s not the difference between CPU and GPU,” said Laanait. “The entire computational model is different. The part that’s relevant is, in a classical computer, if you increase the number of bits by a factor of 10, the amount of information you can process increases by a factor of 10. In quantum computing, it increases by 1,000, and it increases exponentially with the number of quantum bits.”
According to Floether, that’s the reason why there’s such excitement around the technology: Quantum is the only computational model that can be exponentially faster than classical computers.
“The journey is a continuous one,” he said. “Considering that this is such a fundamentally different technology, it requires time to build those skills, build those solutions and get into a quantum state of mind.”
A sign of growing maturity in the field, they said, is that major companies and smaller players alike now have road maps. Intel, Microsoft and IBM are some of the heavy hitters with quantum plans. They’re planning to scale the technology. IBM in particular has hit every one of its milestones and is projected to have a 4,000 qubit machine in the coming years.
“These machines are so complex that you cannot simulate them classically,” said Laanait. “They’re already past that threshold.”
At this point, not every problem can be solved in a quantum manner. It’s critical, said Floether, to do careful mapping between potential use cases. Current problems at which quantum computing currently excels include processing data with a complex structure, simulation and optimization.
Where quantum computing can really shine is in kernel-based machine learning. A kernel, a math function applied to data, can allow people to see more structure in their data.
“If you were to project it to an even higher function, you’d see even more structure, even more patterns in your data,” said Laanait. “With quantum computers you can go to a million kernels.”
The software is one thing. But that software doesn’t have much value unless it has the hardware that can run it, and that’s where the technology still has some catching up to do. But as the tech gets better, the data will get better.
To date, said Floether, health data is about 60% to 80% accurate in terms of data classification in classical models. The early results on quantum computing are powerful, showing the ability to outperform classical results.
“Considering the youth of the technology, this is very promising,” said Floether.
Additional developments are needed to match best-in-class machine learning, said Laanait, including larger feature dimensionality and noise resiliency. But he said the healthcare industry is already on the cusp of quantum computing being the mainstay, and the industry needs to jump on the technology as soon as possible.
“Nobody can do quantum computing alone, but you have to start now,” said Laanait.
Article link: https://www.healthcarefinancenews.com/news/quantum-computing-poised-transform-healthcare
healthcarereimagined 