The Economist's Future of Health & AI Summit

Last week I attended two events hosted by The Economist. The first was the Future of Health Europe. The second, the following day was the AI in Health Summit, which explored what a future digitally driven future might look like and the obstacles to getting there.

The speakers collectively painted a picture of health systems across Europe and globally under acute strain, delivering care against a collective backdrop of geopolitical instability, supply chain fragility, demographic pressures, ageing populations, climate stress, antimicrobial resistance, and constrained budgets.

The speakers explored how health systems must adapt in order to survive and absorb these shocks. One speaker powerfully noted..

"Health can no longer be treated as a cost to contain, it must be understood as infrastructure. Just like roads, schools, and energy grids, health underpins the functioning of society, economies, and democracies"

We heard directly from policymakers and economists about the need to shift how we evaluate investment in healthcare from considerations in terms of clinical activity or cost-per-intervention, to its potential for contributing to productivity, wellbeing, and social stability. The Economist launched its Health Dividend Initiative to explore exactly this, highlighting the need for long-term thinking over reactive policymaking.

The health workforce, arguably one of the biggest challenges facing health systems today. How do we reimagine our human capital capacity to absorb such significant changes? How can we best recruit and retain talent in an era of burnout and shifting career values? Some of the solutions included the need to remodel roles and transition traditional roles beyond hospitals, and leveraging the skills within high-street pharmacies to absorb pressure. As one speaker put it...

"We need to stop thinking of workforce shortages as a recruitment problem and instead to consider it a design problem, where the solution is redesigning care around what people can do (with AI and digital support), not what they used to do?"

The NHS Ten-Year Plan was explored as a live case study in how prevention, integration, workforce reform, and digital transformation will reposition the NHS for a more sustainable and future-fit model of care. Yet there are significant gaps that threaten the plans credibility and deliverability, particularly the tension between long-term preventive ambitions and the short-term pressures of acute demand and constrained budgets. Shifting from a reactive, illness-based system to one that prioritises early intervention and personalised care will require new financial models that reconcile day-to-day constraints with longer term structural investments. Technology was framed as a critical enabler of this shift, a means of reducing administrative burden, enhancing clinical capacity, and releasing financial headroom for reinvestment into redesigned care pathways. However, this is far from straightforward. Persistent infrastructure deficits, outdated IT systems, procurement delays, and clinician scepticism continue to hamper progress.

At the same time, speakers shared examples of where systems are already moving in the right direction, accelerated through innovative patient-centred digital design, and with robust innovation-led leadership, presenting a vision of what is possible when strategy, organisational culture, and technology align.

The topic of health inequality and structural exclusion was discussed repeatedly. The challenges posed to some sections of society showing up in gaps in access, outcomes, and in digital readiness, as well as more broadly in terms of structural exclusion. The promise of community-based care, workforce design of non-traditional roles presented opportunities in bridging these access gaps, but without tailored training and digital support and enablement, new AI digital service models pose risk to further entrenching inequalities in care quality and access. The warning was clear:

"AI will benefit the systems that least need it, and bypass the populations who most do"

As you might imagine, AI was heralded as the future and means through which we can address the challenges plaguing health systems globally. Rather than an ethereal future, speakers described how AI is already reshaping clinical practice in tangible, measurable ways and delivering real‑world impact. Many of the most compelling AI examples came from clinicians and organisations focused on narrowing clinical risk and surfacing unseen signals in patient data.

• In radiology, AI is being used to support mammogram and emergency X‑ray interpretation, reducing false positives and helping clinicians manage increasing volumes.
• In surgery, AI-driven decision support is enhancing precision by identifying tissue boundaries during robotic procedures reducing surgical complications, improving recovery rates and reducing time in hospital.
• In nephrology, a predictive model is being used to flag home dialysis patients who are at risk of dropping out, allowing timely clinical interventions to retain people in their care.
• In orthopaedics, AI tools are identifying which patients are most likely to experience pain after surgery, which enabled more realistic goal-setting and provision of post-operative support.
• In cardiovascular care, machine learning is being embedded onto electrocardiograms and wearable devices to detect signs of heart failure well before clinical symptoms appear.
• Automated documentation tools are generating clear, patient‑friendly consultation summaries that integrate directly into health records, easing clinical burden and improving communication.

These examples aren't prototypes, they are live, embedded technologies already showing clear clinical and operational benefits.

There was plenty of refreshing honesty about failure. We heard from digital health leads struggling to scale tools across fractured NHS infrastructure. AI pilots are everywhere, but most don't scale. That's not because they're bad ideas but because they are hampered by outdated procurement frameworks, fragile IT systems, and clinical workflows that were not designed for automation. One hospital leader described how it takes 10 minutes to boot up a computer on some wards, against that backdrop, promises of 'AI-enabled decision support' will drop down the priority list.

Organisational culture can block adoption. Clinicians need more than regulatory approval and the message was clear regarding clinician's need to understand, test, and slowly build trust in the tools they're being asked to use.

Many speakers acknowledged that without clean, well-structured, longitudinal data, and the governance mechanisms to use it ethically, the success will be constrained. Several hospital have invested heavily in building "AI-ready" infrastructure: annotated image banks, standardised clinical pathways, and curated datasets that allow tools to be validated, updated, and redeployed. But this is still the exception and for too many health systems, clinical data remains locked in PDFs, scanned letters, or non-interoperable platforms.

One of the most powerful reflections came from a German health system leader, who noted that:

"Innovation cannot live in a vacuum, it needs plumbing."

What he meant was that unless we invest in the digital and operational infrastructure that allows innovation to be integrated, we will remain stuck in a cycle of promising pilots that fail to scale.

There were discussions on the need to reform how AI is governed. There was real concern that existing regulatory frameworks, particularly in Europe, are too rigid, too slow, and too expensive for iterative AI tools that evolve as new data comes in. Several speakers called for risk-tiered regulation, where low-risk tools (like triage or admin support) are fast-tracked, while high-risk applications (like diagnostic decision-making) are subject to more rigorous scrutiny. Without this flexibility, many startups and SMEs will either avoid the UK and EU altogether or be forced to scale only through partnerships with better-resourced incumbents.

There were many ethical questions posed:

• What happens when AI identifies a clinical risk, but no one acts on it?
• Who is accountable if a clinician overrides an AI tool and gets it wrong, or follows it and causes harm?

"We must prepare for a world where not using validated AI could be as professionally risky as using it inappropriately"

The challenge will be finding the balance between automation and autonomy, ensuring that AI tools are seen not as replacements, but as aids to human judgement.

Despite the complexity, there was a sense of pragmatism and realism. AI won't magically solve the challenges of underfunding, workforce gaps, or ageing populations. But there was a clear consensus that it has a role to play, if we choose to embed it wisely, govern it transparently, and design it for real-world complexity.

Ultimately, the future of health will be shaped by the decisions we make in the next three to five years. AI holds enormous promise, but whether we can leverage its benefits will be shaped by whether we can govern, embed, and scale these new technological capabilities and actively shape the conditions under which it is deployed and where the need is greatest. Investing not only in tools, but in the foundations; the redesigned AI-driven pathways, the data networks, interoperability standards and federated data approaches, the procurement models and cross-sector partnerships, and the governance and regulation that make those tools usable and useful and safe if AI is to operate across the full complexity of patient journeys.

"We need risk‑tiered regulation where low‑risk tools like triage or admin support should be fast‑tracked, while high‑risk diagnostic applications are subject to more rigorous scrutiny."

We must build, starting with low-risk, high-benefit applications that can demonstrate real impact and earn public and clinical confidence. At the same time, we need to create space for innovation by decommissioning low-value processes, outdated tests, disjointed workflows, and legacy technologies that distract from what matters.

The next chapter of healthcare will be shaped by our choices. We can choose to patch legacy systems with short-term fixes, or we can take a more ambitious path, embedding digital intelligence, redesigned care models, and putting patients at the centre of decision making. AI is not a silver bullet. It's a lever, and like any lever, it needs the right pressure, placement, and purpose to move the system forward.