AI proliferation and 'scale' - healthtech's missing enabler.

AI proliferation and 'scale' - healthtech's missing enabler.

If you're actively following and playing with new Generative AI tools then you'll likely be feeling like you're in the most exciting playground possible, and watching an incredible phenomenon happen in society:

The positive disruptive promise of digital, through the introduction of a General Purpose Technology (GPT).

The hawkish observer will note that OpenAI's GPT (Generative Pre-trained Transformer) shares the same acronym.

In the wonderful book: The Second Machine Age, which to me is THE essential read on the impact of AI in society, they define the power of General Purpose Technologies:

"with their verbal flair, economists call innovations like steam power and electricity 'general purpose technologies'. Economic historian Gavin Wright offers a concise definition "deep new ideas or techniques that have the potential for important impacts on many sectors of the economy". "Impacts" here mean significant boosts to output due to large productivity gains. GPTs are important because they are economically significant - they interrupt and accelerate the normal march of economic progress".

Furthermore they define that the biggest and most impactful GPTs are where there are more applications of 'recombinant innovation', derived from the concept of "Recombinant Growth" by economist Martin Weitzman, through considering the amount of contexts that are able to be [positively] disrupted by the application of a technology.

Naturally the other factors that would need to be considered are the speed that an individual context that can be innovated and re-innovated, and the speed that a technology can develop.

Context - where something happens. e.g. a maternity
service (or part of).        
Technology - something that can be applied to a context.        
Context + Technology (applied) = innovation or disruption
to update that context.        

Whilst the speed of replacing steam and electricity was slow, expensive and offered relatively marginal gains, the phenomenon of AI is very different: the provider can instantly, globally, repeatably update and evolve the technology at pretty much zero cost.

The difference between ChatGPT 3.5 and 4, or Midjourney 6 and 7 etc. can happen overnight with one update on a few central servers.

Which is why many days our jaws hang loose at the fact we can have a week of stunning updates that we can actively use and apply within a few days.

This is where the recombinant innovation takes place.

Immediately people across the globe to set to work to either apply or re-apply their innovation, using generative AI as the general tools to update a context, and then another and another.

The fact that as I write this many of you will be playing with tools, and looking at where you can use it, not just big companies, but individuals like me and you, for an incredibly small cost. Not just playing but building something, and building on things that others are building too. We have the ability to build, and build together at a rate that is wider, faster, more combinatory or re-combinatory than ever before.

And so proliferation happens.

What about healthtech?

Given many of you readers, if not all of you, will be interested in health, many of you will be thinking of applying these new AI tools in health.

However, unlike the open market where we can build rapidly together in various contexts we are subject to a different dynamic:

  1. We don't own the context (place of application) it's held by systems and providers who are keen but largely guard the context from risk and unsafety. So whilst the technology iterates the amount of areas where it can be applied is limited.
  2. The current IT that is used in these areas is not like the technology we use outside. There are big walled gardens controlled by incumbent companies, who don't let us play with them. Many are still based on local semi-cloud versions and differ hugely from provider to provider. So our ability to combine and re-combine is slowed and stifled, in some cases stopped.
  3. Our contemporary systems are under huge pressure, and whilst "significant boosts to output due to large productivity gains" are sorely needed, the most promising technology is not prioritised in comparison to how it is being prioritised and applied in the world outside. So, again, the context in which to apply is reduced, and those seeking to innovate do so OUTSIDE the context.
  4. There are less functional places where those in the context and those building the technology can actively collaborate and work together in order to combine [tech + context] and then recombine - across sites, contexts, and technological breeds. And so the act of enabling the combination and re-combination is stifled, reducing the speed of recombinant innovation.

This is not just the NHS, but there are certain dynamics in the design of the NHS that seem to increase the four below. There are also dynamics that could make proliferation much easier, which is why so many outside the NHS considering it at face value see it as attractive.

Right now time and again I'm hearing people saying that when it comes to healthtech "the biggest problem we have in the NHS is difficulty to scale".

Firstly, that's a statement of circumstance and not the inherent problem.

So what is the problem?

We lack the necessary enablers, underpinned by the right mechanisms to drive the dynamics of recombinant innovation. Period.

So let's look again at the four above and try and define them as NHS problem statements:

  1. The NHS has no clear mechanism to expose access to operational 'contexts' in a way that is safe and open to apply new technology.
  2. The NHS is held hostage by outdated monopolies or duopolies of outdated technological systems, who bar innovation that isn't their own, and in their own are slow and innovating at the speed of their own clunky bureaucratic silos. We're even doubling down on these systems.
  3. We are simply not making technological disruption an ambition alongside (largely analogue) integrative efforts. We have a flawed view that we'll do the reform and then apply the technology, even if fundamental aspects of the reform are predicated on things that we don't have and can't obtain (e.g. workforce!).
  4. We have a cold war mentality between the NHS and industry, where the former is culturally cynical and guarded. Furthermore we have underinvested in peer communication and collaboration systems and tools that can be used across the NHS (NHS Futures is not the answer). Look at how impactful this has been at Buurtzorg!

On the final point I believe down to my bones that the NHS should not be privatised, however, it is an inevitability that technologizing healthcare provision globally is effectively privatising more and more of healthcare provision, since it will be the private sector who will build transformative technology.

Ideally there will be a financing model that tries to guild 'state stakes' as Mariana Mazzucato, eloquently endorses.

Sum

We talk a lot about digital transformation, and digital skills in the NHS. The term annoys me if I'm honest, even if I have to work with it. Because it inherently misses the wood for the trees.

What I would like to see is a genuine, considered mode of thinking that looks at digital dynamics and enablers of recombinant innovation.

If we want a digital revolution in healthcare that drives outcomes and benefits for those receiving care and those providing it, then we can't treat it as a closely controlled and defined thing. We need to open ourselves up to, and embrace the dynamics of technological revolution and find ways to reduce or remove friction.

Then we can proliferate, and scale.

Until then we're talking about the wrong things IMHO.

------------------------------------

p.s. quick ask - if you genuinely found this article helpful or interesting then please do like, comment and re-share - it really helps me to with my big hairy mission.

I put a lot of time and energy into creating the best content for my tribe that I can, instead of going for the other cheap tricks on LinkedIn. So if you genuinely found this article helpful or interesting then please do like, comment and re-share - it really helps me to with my big hairy mission. Thanks in advance. Liam

p.p.s. Now over to me talking about me in third person.

----------------------------

Liam Cahill is a trusted adviser to frontline providers and national bodies on all things digital, with nearly two decades experience of doing tech stuff in the NHS. He has mentored and advised some of the best known names in Healthtech, and they've usually said some nice things about his work . He regularly posts content, ideas and advice on LinkedIn. Check out his other numerous articles and videos here , and subscribe to his LinkedIn newsletter here .

Anna Dijkstra

Director | UK Healthcare at Microsoft

8 个月

Thanks Liam Cahill, there's a lot in here which resonates with our experience as well. That said, my lens is equal part NHS and behemoth alongside healthtech so I'd add here: 1. The NHS is experimenting too, unhindered by their own context. Don't get me wrong, they're hindered by a lot else - time/finances/complex stakeholder dynamics/skillsets. Use cases are in production now. Back to the scale problem: NHS Trusts will rebuild use cases unless we have a way of nationally sharing code. 2. The oligopolies are innovating too. Failing to do so will pose an existential threat to be the most embedded of these suppliers because of the sheer scale of value which can be added to a product in months/years - and I don't think we've seen this in the past. 3. Use case parties are finally bringing traditional (process) transformation teams and IT together in a way we'd hoped would happen for RPA/low-code and other tech innovation capabilities. It's not a solution to (3) and (4) on your problem statement list, but it's a small improvement.

Oliver Smith

Catalyst for Transformational and Responsible AI | Health Strategist | Policy Innovator

8 个月

Really love the analysis of the barriers Liam Cahill, and particularly that on culture, although also with a shout out to the weird centralised / localised dynamic of micro-management and macro-blindness that the NHS does so well. I'm less keen on the idea of making technological disruption an ambition, but I don't think that you mean it as an end in itself. More that, we need to be open to disrupting our ways of working in order to get the better outcomes that we all want, and thinking about the role of new technology in that from the outset is important.

Hi Liam - everything you've written above is bang on in terms of market - the dislike of private sector and the repeat purchases of the same tech is excruciating to watch! - and actually I don't think it changes much in the US either. There, I was told, even if the hospital is private, IT is seen as an overhead - hence, it gets no love. But there's another aspect I have seen as well - no data scientists in hospitals, and IT teams that either build crude applications, or focus on integration only. That will be different in a few research hospitals I guess, but overall I'd be surprised if there were many exceptions (though happy to learn). For the reasons I mentioned, it's difficult for a health delivery organisation to actually appreciate the value and utility of data and the related service improvements that can come from it. However, most of the world's market is still wide open, and I think you'll see innovations proven in other markets, entering (back) into the NHS based on the track record, especially if they are launched by ex-NHS clinicians and staff.

Jason Roberts

Connecting and collaborating to support health innovation

8 个月

Another enjoyable read Liam. Thank you for the effort you take to articulate these points clearly. One item I'd add is the poor health of the data that sits inside the 4 walls you describe. Both the quality and the way data is curated has the power to obstruct and upset progress. Frankly, in many cases you wouldn't want recombinant innovation learning from the information foundation that currently prevails. Garbage in - garbage out is still very much there in the mix.

Damn, you're making me think deeply again Liam Cahill! My initial reflection is that you article approaches this from the lens of an EPR-centric eco-system . In reality Gen AI will tackle the back office, mid office and front office technologies and not just in the EPRs. We see workload diversion using increasingly capable chat bots release human time and improve care across lots of places. I've recently had several conversations about whether a chat bot will replace the intranet. Whether we need web sites as we currently know them. If Gen AI provided a chat interface to near-real time databases, can we reshape and reskill our teams of data analysts to help end users interpret data and deliver decisions and actions? So, I think my conclusion is that its not really a question of 'contexts'. It's a question so culture and leadership. "Who will make the technology driven change happen in your organisation?" becomes the key question.

要查看或添加评论,请登录

Liam Cahill的更多文章

社区洞察

其他会员也浏览了