#48. W/c 8th September '25
Alastair Allen
Partner, Head of Healthcare Technology at EY | Board Member at openEHR UK | Health and Social Care Council Member at TechUK
By popular demand, the AI-generated image of me makes its return this week. After a short break, a few of you reached out asking where it had gone—so it’s back to accompany the lead story in this week's edition of What I Read This Week.
This week in my newsletter, I examine a pioneering whole-slide foundation model for digital pathology, discuss the challenges faced by healthcare professionals when administrative tasks compete with patient care priorities. Additionally, I highlight the crucial role of dental care in overall health and review diagnostic software that significantly increases the rate of full recovery in stroke patients.
A whole-slide foundation model for digital pathology
An article published in Nature discusses the Prov-GigaPath model, a new whole-slide digital pathology framework developed using real-world data from the Providence health network in the US.
Digital pathology poses unique computational challenges, as a standard gigapixel slide may comprise tens of thousands of image tiles. Prior models have often resorted to subsampling a small portion of tiles for each slide, thus missing the important slide-level context.
The Prov-GigaPath model was trained with 1.3 billion pathology image tiles from over 171,000 slides, originating from more than 30,000 patients across 31 tissue types. The data significantly exceed the size and scope of previous datasets such as The Cancer Genome Atlas (TCGA). The model utilises a novel vision transformer architecture, adapted for ultra-large-context modelling by employing the LongNet method.
Prov-GigaPath demonstrates superior performance across 25 of 26 pathology tasks, improving significantly on 18 tasks over previous models. It shows promise in vision-language pretraining by incorporating pathology reports, potentially enhancing pathology diagnostics by enabling more personalised cancer treatments and health care delivery.
Overall, this study illustrates the potential of foundation models to revolutionise cancer diagnostics and could serve as a standard for future developments in the automated analysis of high-resolution digital pathology slides.
I am an oncologist. Please don’t ask me to write a business case
In a Guardian opinion piece, oncologist Ranjana Srivastava argues for the critical importance of dedicated nursing support for elderly cancer patients, noting that age-related frailty, cognitive impairment and social isolation make geriatric oncology a field of growing urgency. Despite launching a clinic within her hospital to address these needs, she found herself hampered by the requirement to write a business case in order to hire even a single nurse—an administrative burden she felt was at odds with her clinical role.
The business-case process proved daunting, drawing on unfamiliar financial projections and hospital bureaucracy rather than frontline patient care. Srivastava documents how, despite producing a proposal that showed a small profit and clear potential to reduce hospital admissions and improve patient outcomes, the plan ultimately failed amid management changes and shifting priorities. She reflects on how the effort diverted her time and energy away from treating cancer and caring for patients, instead forcing her into the role of amateur healthcare economist.
This piece is a powerful reminder that healthcare innovation isn’t just about new approaches to digital, data or technology—it often falters on the simplest things, like giving a clinician the nursing support she needs, or too much bureaucracy undermining the care providers hope to deliver.
Diagnostic software triples rate of full recovery in stroke patients
A new AI-assisted tool is transforming stroke care across the NHS, enabling patients to receive a diagnosis up to an hour faster than with traditional scans. Since its rollout last summer, the software has been used on more than 60,000 people across 100 stroke clinics, now accounting for half of all stroke diagnoses in England. By accelerating brain scan interpretation, the technology helps clinicians make faster treatment decisions at a critical point when millions of brain cells are lost every minute during a stroke.
The results have been described as revolutionary: NHS analysis shows that the number of patients making a full recovery has jumped from 16% to 48% since the introduction of the tool. Recovery is defined as returning to life with little or no disability after a stroke, and the technology has been shown to triple the chance of this outcome. It has also contributed to a rise in thrombolysis treatment rates, where drugs are used to break down blood clots, from 11.6% last year to 12.3% in 2024–25.
Another great example of AI moving beyond hype into measurable impact, with faster diagnosis directly translating into dramatically better outcomes for stroke patients
10 Hard Truths About UK HealthTech & AI
Sabine Azancot sets out ten “hard truths” about UK healthtech and AI adoption, arguing that many innovators misunderstand the system by focusing too heavily on NHS procurement. She highlights a fundamental paradox: while the NHS funds care, real adoption lives or dies with clinicians and patients. Procurement may unlock scale, but true demand comes from the frontline, and history shows even well-funded pilots fail if workflow fit and patient engagement are ignored. Early-stage startups face an impossible bar — expected to show large-scale evidence before they have the chance to survive — leaving the system tilted toward incumbents.
She identifies deeper adoption gaps, from sidelining patient engagement to vague visions of neighbourhood hubs that risk fragmentation without anchoring in GP and PCN cores. The lack of clarity in public–private partnerships further slows innovation, especially around prevention, where agile commercial models and data integration could make the biggest difference. Azancot argues that innovation should simplify, not add layers, warning that piling platforms and portals onto overstretched staff only entrenches inefficiency. She also cautions that AI oversight is lagging: without real-time vigilance and governance, safe tools may stall while risky ones proliferate.
Looking globally, she points to lessons from France, Germany, Catalonia, Singapore and Estonia, which demonstrate both pitfalls and proven enablers for scaling innovation. Her final argument is stark: unless the UK rebalances investment upstream into primary and community care, it risks another decade of pouring money into hospitals while symptoms worsen. With unique assets like a unified NHS and rich longitudinal data, the UK has a chance to lead — but only if it couples speed and openness with clinical vigilance and upstream funding.
A Healthy Mouth for a Healthy Life
An article on Digital Health Buzz makes the case that oral health is fundamental to overall wellbeing. It highlights how poor dental hygiene can contribute to gum disease, tooth loss, and systemic conditions like heart disease and diabetes, while also affecting confidence and quality of life. The piece stresses the importance of daily habits—brushing, flossing, diet choices—alongside regular dental check-ups and lifestyle changes such as quitting smoking and managing stress.
It was good to see the NHS 10-Year Plan also recognises the need to reform and strengthen dentistry, with commitments to expand access, focus on prevention (particularly in children), and shift services “from analogue to digital.” As those reforms take shape, digital tools—such as teledentistry, smart oral health devices, and integrated digital records—will be key enablers of a more accessible, preventive, and equitable dental system
Tech & AI Business Leader | European HealthTech | Strategy • Product • Growth
1wGrateful to see the ‘hard truths’ of UK's HealthTech & AI adoption getting more visibility — thanks Alastair Allen for picking it up in this week’s newsletter
Honorary Professor in Engineering & Applied Science at Aston University
1wThanks for sharing. I can certainly relate to the Business case oncologist.