Return migrant on the verge of something new

Juha Merimaa

13.6.2024

Olli Kallioniemi, Professor in Molecular Precision Medicine, is working on a long-term research project aimed creating a data-based way to diagnose acute myeloid leukaemia, a cancer of the white blood cells, more accurately using artificial intelligence.

It’s an ambitious and innovative project.

But Kallioniemi doesn’t really want to talk about it. A single study is only a fraction of a major change.

“Artificial intelligence will revolutionise the entire field of medicine, from public health to diagnostics and treatment decisions,” Kallioniemi says.

“This is the big picture that I want to highlight as a researcher. None of us yet knows what will change and how, but it’s already certain that many things will change. So now is an important moment to look ahead.”

He emphasises that this is not hype. Advances are currently being made at a breathtaking pace.

”It’s estimated that in 2001 a thousand dollars would buy the equivalent computing power of an insect brain. In 2010, the brain of a mouse. Today, we’re already in a situation where a thousand dollars is enough for the computing power of the human brain. In a few years, we will be in a situation where a thousand dollars will provide computing power equal to the collective brain power of all humankind.”

Artificial intelligence will revolutionise the entire field of medicine, from public health to diagnostics and treatment decisions

Kallioniemi

World’s best registers

Computing power is not enough, however. Artificial intelligence also needs data to support it. This is where Finland is one of the best places in the world.

For the past nine years, Kallioniemi has worked as Director of the Swedish Science for Life Laboratory. He’s now returning to Finland with the help of a joint grant from the Finnish Medical Foundation, the Sigrid Juselius Foundation and the Jane and Aatos Erkko Foundation under the Brain Gain Programme. The five-year €2,500,000 grant will allow Kallioniemi to establish a research group in stages during 2024-25 at FIMM – Institute for Molecular Medicine Finland at the University of Helsinki.

“Although research is not funded as much in Finland as in Sweden, Finland’s advantages include much more accessible research data and registers. We’ve already been working systematically for a decade to ensure, for example, the availability of biobanks, genomic data and various healthcare registers for research. In a secure and ethically sustainable manner.” 

This gives researchers access to a huge amount of health data, including information on the various risk factors that predispose people to different diseases. Sometimes information can come from an unexpected direction. Kallioniemi mentions, for example, fundus images, which are routinely taken when trying to screen for eye diseases. With the help of artificial intelligence, fundus images have also unexpectedly made it possible to predict the long-term risk of heart disease and neurological diseases.

How does artificial intelligence determine cancer?

The use of artificial intelligence in medical research is nothing new. For example, artificial intelligence can help radiologists screen lung images for tumours and pathologists to interpret histological sections.

A good start is still not enough, Kallioniemi says. “In the first phase, artificial intelligence has been taught to diagnose the same findings as doctors. It does this faster, more cost-effectively and sometimes even more accurately than a human. Still, it’s just a matter of mimicking humans, and humans need to check the results provided by AI.”

Kallioniemi hopes to get artificial intelligence to diagnose things in a new way from the start.

“Just think about how cancers are identified,” Kallioniemi explains.

Pathologists have certain criteria that they use to classify which tumours are malignant and which are not, or how aggressive the cancer looks.

“But is this the most accurate set of criteria possible? Could AI do it in a whole new way, perhaps differently and, in time, better?”

In Kallioniemi’s vision, AI will no longer be given human-generated criteria, but a huge amount of information, such as genes, laboratory results, medical imaging, disease prognosis, treatment responses and so on. With this information, AI could itself create criteria for new diagnostics and classification.

Kallioniemi believes that AI can, for example, detect things in image data that people miss.

”The human eye is not very good at measuring things either. Instead, AI calculates distances from images, for example, and determines features with pixel accuracy,” Kallioniemi points out.

”I would compare the difference to watching and old black and white TV or an HD colour TV. You get a lot more information from one of them.”

As such, the situation where AI perceives things better than humans is not new. In Go, for example, AI beats humans but occasionally makes moves that seem absurd to us, but whose cunning is only revealed after a long time.

The utilisation of AI in medical research is also a broad and diverse field, and a traditional professorship in medicine at a university does not cover such a job description. The grant will allow for transition and, with time, the creation of a new role.”

Brain Gain enabled return

Kallioniemi’s return to Finland is due not only to research opportunities but also to personal reasons. His wife and adult children live in Finland.

”I’ve spent about half of my research career, 20 years in total, abroad, in the United States and Sweden. At this stage, returning to Finland seemed to make sense.”

He thanks the Brain Gain Programme grant for making his return a reality.

“The utilisation of AI in medical research is also a broad and diverse field, and a traditional professorship in medicine at a university does not cover such a job description. The grant will allow for transition and, with time, the creation of a new role.”

Kallioniemi’s future supervisor, Professor Jari Koistinaho, Director of HiLIFE – Helsinki Institute of Life Science, is also excited about the researcher’s return.

“Olli has previously worked for a long time in our FIMM unit. We know him as a creative and transversal researcher. I think it’s great that he’ll be able to move away from administrative tasks and back into research.”

Koistinaho thinks the University of Helsinki is a good place to study the possibilities of AI.

”We have both register and artificial intelligence expertise”

Tumour examined by AI

But how about a brief summary of the research? Kallioniemi assures us that he will also continue his cancer research. The idea is also to use AI, but this time specifically in the development of cancer diagnostics.

“As such, the idea is exactly the same as in the bigger picture. We’re taking tissue samples from decades ago, digitising them more accurately and trying to teach AI to diagnose tumours and distinguish the dangerous ones from the benign ones.”