Is the tide turning on artificial intelligence (AI)? The technology has gone from being viewed with suspicion, a potential risk to everything from job security to personal data, to a potential helping hand in the workplace that will augment human effort rather than replace it.
But healthcare is still an underserved area when it comes to the use of AI. Things initially looked promising. More than a decade ago, IBM’s Watson was expected to usher in an era of AI-powered healthcare that would help diagnose illnesses more quickly by leveraging the power of the new technology, and offer up more effective treatments in a shorter time frame.
But the promise was short-lived. Despite stories about Watson identifying rare cancers in a matter of minutes and helping to put people on the correct treatment path, the technology had its problems. Some doctors complained that it was recommending treatments that simply were not available, having been trained on data available in certain regions with different healthcare systems.
Watson was eventually sold for parts, becoming a footnote in tech history.
The new generation of AI, however, could be more useful. Recent research projects found AI could be used in different areas of healthcare, from taking the pressure off overworked X-ray technicians to helping identify breaks and fractures more accurately and examining the brain scans of stroke patients to help identify the timescale of a stroke. Another AI tool was found to be proficient in predicting the chance of developing diseases such as COPD, kidney disease or Alzheimer’s, and other software was able to detect almost two-thirds of the epilepsy lesions that human doctors had missed.
[ No escape for professions from rolling revolution that is artificial intelligenceOpens in new window ]
For Irish company OpenSky, AI has been a game-changer for its work. The Naas-based technology company, which was founded in 2004, is more known for its digital transformation skills, but in recent years it has been building a healthcare and pharmaceutical business that has grown in importance to the company. The company has turned its skills in providing AI-powered IT solutions and consultancy services to both public sector and private organisations to the healthcare sector.
As part of that shift, the company in 2022 decided that it would invest heavily in research and innovation at the company.
AI is also a great leveller, open to companies of all sizes as long as they are willing to put in the work. Investing in research and innovation is crucial for companies that want to take advantage of AI and help transform the healthcare system.
“That’s one of the things that we feel is important to us as a company, even though we’re an SME and we’re not massive,” says Michael Cronin, cofounder and managing director. “But innovation just isn’t for multinationals.”
One of its most recent projects has seen the company deploy AI and data management to help in the fight against HPV-related cancers through vaccination programmes.
According to the World Health Organisation, cervical cancer is the fourth most common cancer in women globally. In 2022 there were 660,000 new cases, with about 350,000 deaths from the disease.
But it is preventable and curable, once it is detected early and managed effectively. The WHO targets around this, seeking to increase vaccination rates to 90 per cent, boost twice-lifetime cervical screening to 70 per cent, and treatment of pre-invasive lesions and invasive cancer to 90 per cent.
The VaxLens service, which was rolled out in partnership with a global pharma giant, is playing an important role in that, helping to tackle early detection of the cancers across national HPV vaccine programmes around the world. OpenSky’s platform provides real-time, AI-driven insights using anonymised data on demographics, vaccination status, socio-economic indicators, healthcare system statistics, screening rates and treatment outcomes.
“The importance of monitoring and surveillance in vaccine programmes is crucial, because, if you’re investing in a national immunisation programme, you need to know the efficacy of the actual roll-out itself,” says OpenSky cofounder William Flanagan. “What’s the market access like? Are there disadvantaged areas? Are people getting the vaccine in a timely fashion when they need it and so on, and to be able to monitor that on an ongoing basis. We’re able to do that in Ireland here, because there’s a database here that tracks that on an ongoing basis.
“However, what was surprising to me was that that’s not the case in a lot of areas around Europe. We worked with MSD to figure out how we can move the needle in some of these territories that don’t have that type of monitoring, that aren’t publishing data, that aren’t actually monitoring it in real time.”
The company initially looked at other areas in Europe and how data around vaccination programmes was being collected and used, before coming up with a proof-of-concept study using data available from Denmark.
The idea is to bring all the publicly available data together and use AI to drive insights and figure out what is most likely to help drive vaccination uptake in different regions, and by extension, help cut the rate of cancers from HPV.
Generative AI may be an attention grabber, but artificial intelligence itself has been around for decades.
“AI is not necessarily new; it’s the enablement of AI over the last few years that is the big game changer,” says Pedro Varela, chief AI officer of BJSS. “It’s the computational power; it’s the fact that now we can use cloud systems like AWS that just facilitate all of these things.”
Much of the benefit of AI comes from the algorithms being faster to process vast quantities of data compared with human counterparts, and never needing to take a break.
Without patients knowing about AI, it makes it very difficult to explain what we may be doing
— Rachel Smith of Beacon Care Fertility Group
Private healthcare company Beacon Care Fertility Group has begun using the technology in its fertility services, training the programmes on the previous work of its embryologists to help lighten their workload.
“There’s been an explosion in AI in all areas,” says Rachel Smith, the company’s embryology innovation and AI lead. “There isn’t probably an area that isn’t being touched from perhaps prediction – using big data sets to understand the patient’s chance of success initially, before they even start treatment – to how we tailor their treatments to their specific circumstances to get the best possible outcomes, and then all the way through to examining eggs or sperm using AI.
“Understanding which embryo has the best chance of implantation from a group is also the most important decision that an embryologist has to make. That is a difficult decision to make with just looking at the appearance of an embryo, so we use an AI tool to help inform us in greater detail about the embryo. Using time-lapse information, we can analyse thousands of images with AI to understand the potential of that embryo, supporting our embryologists in making the correct decision on which embryo to replace, and also helping the patient to understand important decision-making with more information at their fingertips than they had previously.”
[ Does AI have a role in helping patients get more out of healthcare?Opens in new window ]
At the heart of it is deep learning, a subset of AI techniques that allowed Beacon Care’s technical partner, BJSS, to try to infer patterns from images.
It helped that Beacon Care Fertility had built up a huge library of images over the years, annotated manually by embryologists and creating a high-quality data set. That could subsequently be used to train AI to work in tandem with the human staff to cut down on time spent on manual tasks. The tool has seen more than 63,000 embryos; a busy clinic might see 10,000 over a decade.
“Those two things, the images and those manual annotations, has allowed us to train the AI because we had quality data to be able to partake in this relationship with BJSS and create something amazing for our patients and our staff. They don’t have to spend hours and hours of time manually on annotating embryos, which would take about 20-30 minutes per patient. So this is a great time-saving tool as well for us to use in-house.”
Trust remains a big issue in implementing AI, regardless of sector. But Beacon Care says its use of the technology protects the privacy of their patients, with the in-house development of the tools the organisation uses giving it more control over data use.
The growth in popularity of generative AI has also made it easier for companies to introduce the high-tech techniques.
“Without patients knowing about AI, it makes it very difficult to explain what we may be doing, or how we may be pushing the envelope slightly on what we would like to do in the future as well,” says Smith. “Having it more socially acceptable helps.”
