AI advances from the lab stage to practical medical applications

Artificial intelligence has the power to transform healthcare, and the shift is only beginning

Professor Tiziana Margaria, chair of software systems and head of the department of computer science and information systems, University of Limerick

If the promise of artificial intelligence (AI) is to mean anything, it has to offer us more than better shopping recommendations or targeted advertising. In fact, scratch beneath the surface and you will find it already is.

Professor Tiziana Margaria, chair of software systems and head of department at the department of computer science and information systems at the University of Limerick, said that AI has earned a place in industry.

“It is picking up in the sense that the applications are really in use. Big data is becoming pretty popular,” she said.

While we tend to think of AI in terms of the web, in fact it has found a home in the material world. The manufacturing sector, Margaria said, was a keen user of the technology, with a lot of work going on in the national Advanced Manufacturing Centre (AMC) as well as in private industry.

“There is a tendency to data-driven optimisation of production processes,” she said.

“In addition, we see it in assisted driving, which is where things stand rather than fully autonomous driving, where AI is used to prevent accidents”.

Margaria, who is also vice-president of the Irish Computer Society (ICS), said that one of the most interesting areas was the application of AI medicine and healthcare.

“There are a lot of applications in healthcare, of course, such as image recognition,” she said.

In this, the objective is to assist clinicians, not to replace them or de-professionalise their jobs.

“One of the objectives is to increase the precision of diagnosis. It’s about assisting the doctors; AI is not there to replace the experts, but it can do pre-analysis as a form of screening,” she said.

Margaria’s colleague, Dr Nikola S Nikolov, senior lecturer at the UL department of computer science and information systems, said use of AI was only going to grow as time passed.

This was not only because the capabilities of the technology were growing rapidly, which is the case, but also because existing applications had not yet worked their way into common clinical use.

“Right now, it’s cutting edge. It’s not as common as it could be,” he said.

Nikolov, who also serves as the course director of the MSc in artificial Intelligence and machine learning at UL, said that the first port of call for AI was in diagnostics, specifically image analysis.

“In healthcare, one of the applications is for automatic detection of objects and images. This is useful for diagnosis of various diseases, looking at MRI [magnetic resonance imagining] images, for example”.

Examining and sorting images is unlikely to be where it ends, though. In fact, where the technology will go is not yet clear, he said.

“Based on how much progress we had [in AI] in the last ten years or so, it is quite hard to predict how far it will progress, as we have seen so much,” he said.

Other possible diagnostic uses include recognising rare diseases, which by their nature may not be otherwise considered.

However, diagnosis is not where AI’s role will stop, Margaria said. Alongside other technologies, including robotic process automation, as well as actual physical robotics, AI is working its way into treatment – again, to assist rather than replace.

“There is an aspect of robotics, for example in terms of microsurgery: controlling high-precision machines from elsewhere in the world. That's very important for accessibility of care, especially now as there is a lack of care everywhere in the world,” she said.

Another role for AI is literature, specifically creating a ‘bot’ that understands the specialist literature and can suggest something to read, whether to a medic, a patient or a family member.

“In UL we are working on a patient information portal. People are exposed to a lot of literature that is not vetted and not advisable, so we are doing an analysis. It could give someone the opportunity to upload a journal or magazine article and have it validated,” she said.

For now, these are early days, whereas areas such as cancer diagnosis and therapeutics are more mature.

“The information portal is newer and itself needs to be validated,” she said.

Margaria said the ICS is very attentive to developments in AI, including by supporting universities and the practitioners such as by creating courses. Indeed, with the European Commission’s 'Digital Decade' agenda for Europe including the target of reaching 20 million IT specialists by 2030, attracting more candidates studying disruptive technologies such as AI will be essential for meeting EU, as well as national, goals, she said.

Also, while AI is relatively novel, Margaria said that it should not be disconnected from either the material world or from the wider fields of computer science and information technology.

“Just AI and ML, they will not solve all of our problems. They have to be connected to reality. The developments in AI and ML support developments in traditional software engineering and vice versa,” she said.