While AI adoption in healthcare has been cautious, the sector is now finally embracing AI technologies at an accelerating pace, particularly in areas such as medical imaging, virtual care, and workflow optimisation.
With its ability to absorb and analyse vast amounts of data from clinical trials, research papers and other sources, the increased use of AI has profound implications for the future of healthcare.
According to Dr Ronan Glynn, partner and health sector leader at EY Ireland, healthcare’s hesitancy when it comes to AI can be attributed to the critical importance of patient safety, regulatory requirements and the need for clinical validation.
“However, these challenges are being actively addressed as healthcare organisations increasingly recognise the transformative potential of AI – and the need for clear governance frameworks to support its safe and effective deployment,” he says.
READ MORE
The HSE, for example, has established a centre of excellence to facilitate the adoption of AI and automation solutions. In tandem, an AI in health strategy is being finalised and Hiqa is developing a national framework to provide an overarching set of principles to guide and promote a responsible and safe approach to the use of AI in health and social care.
With complex pathways, specialised skills and a need for precision, healthcare sets a particularly high standard for AI integration, says Raymond Martin, director, health tech, PwC Ireland.
“The stakes are high – from patient safety to regulatory compliance – requiring thoughtful planning and testing,” he says. “Yet, the potential benefits of AI in healthcare could surpass those in other industries. AI is poised to improve all areas of healthcare within a decade. It will increase the effectiveness of preventive care, forecast health emergencies for people, amplify health coaching, accelerate genomics analysis, revolutionise demand forecasting and much more.”
From a clinical perspective, the focus to date has largely been on the potential use of AI to improve diagnosis, particularly in digital pathology and medical imaging – where it is being used to support identification of cancer, stroke and fractures, and the discovery of new treatments and vaccines.
“AI’s most notable success to date is in radiology, where AI-powered image analysis is now becoming routine in diagnostics,” says Martin. “AI doesn’t replace radiologists but enhances their efficiency, allowing them to see more patients, focus more effort on complex cases, thus maintaining quality care at much lower costs and reducing backlogs at the same time. This model will be repeated across other pathways, and that is where breakthrough gains will come.”
More broadly, Glynn explains that AI holds significant potential to help streamline and automate operational and administrative work – for example, optimising patient flow, supporting staff rostering, creating discharge summaries for patients leaving hospital, supporting the booking/scheduling of appointments and accelerating recruitment processes.
But it also holds significant potential in population health management, helping to identify and support people at risk of illness or injury and allowing health services to intervene early.
“For example, in northeast London, AI-powered prediction software uses routinely collected data to identify those patients who are at risk and who require immediate preventive support to avoid future, unplanned visits to the hospital,” says Glynn.
Excitingly, AI can accelerate every phase in the notoriously slow discovery and development cycle for pharmaceuticals, from earliest research to clinical trials.
“It can also improve the quality of these phases by simulating interactions and adverse reactions, eliminating non-viable options faster, and discovering new indicators of probable success,” says Martin. “We have seen this in action with the development of Covid-19 vaccines, and that is why so many in the industry have hope for breakthroughs in areas where we’ve previously had limited success. It may open new avenues for research into Parkinson’s, motor neurone disease and rare or hard-to-treat cancers, among others.”
“AI has the potential to streamline every step of the clinical trial process while also expanding the reach of those trials, opening them up to a much broader group of participants, making those trials more inclusive, accessible and representative,” Glynn adds.
AI can also be used indirectly to predict the onset of disease at its earliest stages. Cal Muckley, professor of operational risk in the banking and finance area at UCD College of Business, has been investigating self-reported money management difficulty (MMD) as a novel predictor of cognitive decline due to early-stage dementia.
“We thought maybe there’s a silver lining to these financial mistakes and perhaps they might indicate early-stage dementia, and this should be part of the screening process in society to pick up people as early as possible, because being diagnosed early is very advantageous,” Muckley explains.
The results were stark: a high performance AI model using MMD offered “invaluable predictive information” relative to other signs of early dementia such as blood biomarkers.
“There are all these lead indicators of clinical diagnosis of dementia but we’re adding a new one called money management difficulty, and lo and behold, it turns out to be the most important of all of them, except maybe age,” says Muckley.
An opt-in system could allow financial institutions to protect their customers who may be developing dementia and allow them to transfer financial control at an appropriate time.
But while its potential is enormous, Glynn warns that AI is not a magic bullet for all of healthcare’s woes.
“The key when thinking about AI is to focus on solving problems, solving real-world clinical and operational needs rather than simply using AI for the sake of it,” he says. “Instead, it is one more, albeit very powerful, tool to support the delivery of high-quality, safe care in a sustainable way.”
