AI promises to revolutionise healthcare, but adoption often stalls at the pilot stage. Núria Vàzquez explores the promises of AI for the sector, the barriers to real-world deployment, and how to unlock its full potential.
As healthcare data shifts from more structured to increasingly complex and diverse formats, AI holds genuine promise to unlock the next wave of healthcare innovation. Between 2019 and 2022, healthcare was among the top industries for AI investment, receiving US$31.5 billion in equity funding. Major players beyond the health sector – such as Google, Amazon, and IBM – are also heavily investing in this space.
The European Union (EU) defines AI as the “capability of a computer programme to perform tasks or reasoning processes that we usually associate with intelligence in a human being” – a definition that clearly distinguishes it from basic automation processes. The potential benefits of using AI to support decision-making in overstretched health systems are undeniable, but how far has that promise translated into real-world impact?
What can AI be used for in healthcare?
One can easily get lost in the ocean of articles, talks, and technical reports praising AI solutions in addressing existing health sector constraints. Despite this seemingly endless list of applications, AI can easily be grouped into four main categories.
Without a doubt, applications in clinical practice receive the most attention and investment. AI interprets an array of data types (e.g., imaging, -omics, EHRs) to support clinicians with patient diagnosis or treatment pathways. Imaging applications are the most mature, though other areas such as pathology and intensive care are showing promising results. AI is also increasingly applied in biomedical research, including drug discovery, clinical trial design, and digital twins. These applications often provide higher benefit than clinical tools, as they do not require the strict certification process that clinical solutions must undergo.
Less well-known, but highly impactful, are applications that streamline administrative processes. These include support in completing clinical notes, detection of fraudulent activity, and patient flow management. While these tend to receive little press coverage, anyone working in the health field understands their importance. For example, bed availability directly affects the number of cancer surgeries a hospital can perform. The American Hospital Association (AHA) estimated that 40% of non-clinical tasks and 33% of clinical tasks could be performed by AI. With workforce shortages well-documented, unburdening staff from tedious tasks improves healthcare system efficiency.
AI can also support public health professionals in identifying demographics or regions with high disease prevalence or high-risk behaviours. Beyond traditional patient-focused frameworks, AI can integrate citizen data with a broad range of external sources such as news, microbial labs, and mobile devices. This enables targeted health promotion initiatives where they can have the greatest impact.
Stuck in the pilot trap
Despite the enthusiasm, real-world examples of AI adoption remain sparse, and many are indistinguishable from pilots. It is difficult to gauge the maturity of AI deployment – whether a project is an academic experiment or a fully integrated clinical tool supporting daily tasks. Translating research into clinically viable settings remains a significant challenge. A 2024 study by the American Hospital Association (AHA) concluded that, in 2022, only one fifth of American hospitals implemented AI solutions for administrative tasks.
Unfortunately, tools that thrive in R&D environments may struggle in overstretched healthcare systems, where clinicians lack the time, resources, or training to adopt them effectively. All indicators suggest that AI has fallen into the well-known “pilot trap”.
The evolution of digital health in healthcare has been shaped by a continuous interplay between advances in Information Technology (IT), clinical research and practice, along with cultural and structural shifts. This co-evolution has not been seamless; progress has alternated between breakthroughs and bottlenecks, and it is important to learn from them to ensure the promises of AI become a reality.
What does the future hold?
In August 2024, the EU made history with the approval of the AI Act – the world’s first legislation specifically aimed at regulating AI applications. This landmark act seeks to protect EU citizens from the potential harms of AI, while maintaining Europe’s global competitiveness. European Commission President Ursula von der Leyen marked the occasion as a “historic moment,” noting that “the AI Act transposes European values to a new era.”
The regulation will take effect in August 2026, bringing significant implications for the healthcare sector. Under the AI Act, any medical device using AI will be classified as “high risk,” just one tier below full prohibition. To facilitate market advancement by European innovators, the EU has developed supporting infrastructures, including regulatory sandboxes, the AI Office, and the four Testing and Experimenting Facilities (TEFs). Bax is directly involved in TEF-Health, one of these facilities, where 52 partners with expertise in AI and robotics collaborate to fast-track European innovation to market. Our participation gives us first-hand insight into the benefits of these collaborative infrastructures in helping European SMEs secure Europe’s place in the driving seat of AI innovation.
Will this be enough? Drawing lessons from our experience in past innovations (e.g. precision medicine, digital health), at Bax we believe that whilst appropriate governance is key to ensuring AI solutions are safe and add value, other areas require attention in parallel. Firstly, significant investment is needed to improve existing IT infrastructure and interoperability across databases. Secondly, the skills gap must be addressed by training existing clinical, administrative, and IT staff. Too much emphasis is often placed on clinicians’ perspectives, but in complex healthcare systems, teamwork keeps the ship afloat.
Work with us to drive better healthcare
In the Healthcare Innovation team, we believe that healthcare systems are among society’s most valuable assets, and we work to secure their sustainability for future generations. We do so by focusing on three main workstreams:
Tackling bottlenecks: Recognising the limitations of current frameworks, we bring stakeholders together to accelerate the adoption of innovative treatments and technologies, ensuring equitable access to quality care.
Facilitating the adoption of innovative solutions: Leveraging our sector’s expertise and extensive networks, we offer strategic insights into healthcare markets, competing technologies, and emerging trends.
Empowering patients & citizens: Through our experience in Patient and Public Involvement (PPI) and healthcare innovation, we empower patients and citizens to actively participate in the management of their health.
Want to collaborate or learn more? Get in touch.
