Mapping Digital Innovations in the NHS

In this post we attempt to map the ecosystem of organisations centred around digital innovation using a tool called Issue Crawler, which generates hyperlink networks.  
In this post we attempt to map the ecosystem of organisations centred around digital innovation using a tool called Issue Crawler, which generates hyperlink networks.  

We look at the NHS Innovation Centre’s list of organisations and see if they connect to “HealthTech” start ups or local facilitators like Leed’s Ripple. We find an interesting lack of connectivity within the web presence of these organisations and offer some thoughts on why this might be.

A key question driving our investigation is to what extent has the NHS has capitalized, or not, on the apparent promise of ‘big data’ or ‘wearables’, setting aside for the time being the question of whether the hype around these technologies is actually warranted.

Dan Sheldon recently lamented the fact that the NHS has lagged behind other parts of government in embracing technological change despite there being no shortage of frameworks and calls to action. Sheldon, who used to work for the Government Digital Service but recently moved to the Department of Health, feels that one barrier is the legacy of failed data initiatives. The National Programme for IT (NPfIT) back in the early 2000s, for example, was a rather costly false start. More recently, the digitization of patient records though the care.data initiative prompted outcries from concerned citizens and groups like MedConfidential described in another post. However, as Sheldon describes, there have also been some notable successes such as a sexual health app SH:24 and projects like NHS Hack Day.

So why is the implementation of digital data initiatives in the NHS proving so difficult? In the United States, for example, the main barriers to implementing digital solutions at a large scale seems to be the fragmentation of the industry – between individual states, healthcare providers and numerous private insurance companies all with different data standards and procedures. At first glance, the NHS should be a more fertile ground for data based innovations as the primary provider for most of the population but, as we’ll see in this post, the NHS is an immensely complex organisation and set up to deal with particular kinds of innovations as opposed to others. Now obviously many important initiatives like NHS Alpha emerge internally but in this post we are interested in how tech companies or start ups might be involved in this process or not.

To better understand the ecosystem of organisations implementing digital health in the NHS we again turned to digital tools, this time a program called issue crawler which maps networks of hyperlinked websites to visualize the possible relationships between these entities. As the name suggests, the tool was originally used to study ‘issues’: controversies or political conflicts over, for example, climate change or the 2000 American election. By crawling websites related to an issue and the websites they link to, Issue Crawler can show the tensions between different sides of a debate or more simply, who links to who through whom. Although digitization in the NHS is not, for the most part, an outright controversy, this tool can still help us grasp in an exploratory way, the relationship between various organisations as represented by their web presence. Of course many of the key pathways to innovation will be offline and not made visible in this way – something which our ongoing interviews with industry and government will help to bring out – but web mapping provides a very helpful place to start the discussion.

The NIC and the Downstream Model of Innovation

However, before we can do such an operation, we need some starting points. One helpful reference is the website of the NHS Innovation Centre (NIC). The NIC is a portal for individuals and organisations looking to create and disseminate innovative products and services within the NHS and provides a list of organisations which are crucial to this process. We were curious to see to what extent this existing infrastructure for technological change was in dialogue with tech companies and tech hubs working with digital data.

We entered this list unfiltered into the Issue Crawler interface and first mapped only the interactions between these organisations.

When a website (a coloured dot or node) links to another website, they are connected by an arrow. The colours refer to the types of websites (.org, .com, .gov) and the size refers to the number of times the website is linked to by other pages in the network – this could be seen as a rough measure of how important the entity is, in terms of hyperlink referrals at least.

Predictably, the central node of this network is the Department of Health, the government agency in charge of the NHS, flanked by a few distinct branches: academic funding in the form of research councils and charitable trusts (Wellcome, Engineering and Physical Sciences Research Council, Medical Research Council, Bioscience and Biology Reserch Council, etc.), companies and organisations specialising in taking new technology to market (Medilink, Rdforum, Nesta etc) – including those concerned with design, and evaluation and testing (CT toolkit, Evidence NHS and NICE).

These clusters, roughly correspond to the steps identified on the front page of the NIC Knowledge Base: 1) identify the problem 2) design the solution 3) refine the solution 4) evaluate it 5) sell it. This conforms to what we in the sociology of science sometimes call the ‘downstream’ model of science. This refers to the widely held understanding that science works in stages – moving from abstract to more concrete uses. First scientists in the laboratory (or academia generally) develop novel ideas, approaches or findings, then these ideas are gradually filtered down to industry and then everyday use in the form of new technologies. This seems simple enough, and a helpful starting point, but this rather linear sounding process belies some of the complexities involved in innovation.

Firstly, defining a problem is itself no easy task. There is a whole set of literature in the sociology of science about how the act of correctly defining a problem in some sense entails its eventual solution and the parties able to solve it. In this sense whoever is authorised to define the problem has a lot of power. Social scientists have also argued, and the UK government Special Committee on Science and Technology have agreed, that involving laypeople and end users at the beginning of the process of scientific development (rather than as a token gesture at the end) can help offset risks, address ethical concerns and problems of implementation sometimes associated with novel science.

One way in which particular needs and problems can be fed back ‘upstream’ from the clinical setting to technology companies is though public tenders. There is indeed one section of the website in which practitioners can pose everyday problems in need of technical fixes. Input from practitioners and patients are also folded into the process through Wouldn’t It Be Great If (WIBFI) workshops and other events. So one question to consider as we proceed is are the kinds of problems which might be solved by digital data identifiable using the same mechanisms and engagement events?

Another point which becomes immediately apparent when looking at the organisations devoted to refining the design and evaluating successes is that they are very much oriented to certain kinds of innovations: prescription medicines and medical devices – (such as lacroscopic cameras, sensors and handheld tools) rather than IT software, smart phone apps or digital infrastructure for big (patient) data. The primary evaluation body NICE, which must approve all devices and drugs, is not required or necessarily equipped to deal with these technologies as part of its remit.

But even if digital technologies and apps do not need to pass through these gatekeepers, there are still quite a few barriers to distribution. A representative from the Department of Health we spoke to suggested that much technological innovation emerges directly from consultant doctors in closely knit teams in specific hospitals. The problem then becomes rolling these ideas out across trusts (products or services need to be sold individually to local NHS trusts of which there are 100s). Only large infrastructural projects like care.data and the NHS digital ‘spine’ are rolled out centrally by the HSCIC (Health and Social Care Information Centre) but many data driven innovations might require this kind of centralised implementation.

So the existing innovation structure is not particularly set up to accommodate digital data innovations. This gap however is certainly recognised by the NHS and bodies like the Medical Research Council who on their website state:

“The need for partnership working is no more evident than in efforts to harness the huge potential of healthcare, socioeconomic and biological data. We have made great strides in recent years with the introduction of the cross-funder Farr Institute of Health Informatics Research, as well as our investments in medical bioinformatics and ‘omics’ technologies. But there is a risk of a fragmented landscape when so many initiatives and partners are involved..”

There are also number of awards and competitions being put on by these organisations to specifically foster innovations in the area of digital data. The point however is that these sorts of initiative exist on the periphery outside the existing infrastructure for delivering change.

This is something we can see when we use issue crawler to ‘crawl’ webpages outside of this network. The program takes the links provided by the NIC and visits all of the websites they link to and then determines which of these pages interlink. The resulting network shows the relationships between the NHS innovation pages and the wider online neighbourhood.

What is striking about this network is that it does not extend very far beyond the community already defined by the NIC. The centre of the network becomes the academic research councils and institutions which are by their very nature keen to link widely: disseminate findings and attract applicants. But the academy is certainly not the avenue through which wearable and app development is mostly occurring and this list has no links to tech industry contacts or facilitators.

So what does this field look like from the other side, from the websites of successful Digital Health start ups. Do they interface with the NHS?

Lack of Links with the Tech Industry

Next we used a very different kind of list, a tech bloggers’ account of the ‘seven most disruptive health apps’ in the UK which includes Big Health, CareZapp, Doctor Care Anywhere, Trial Reach, Genix and Health Unlocked. Disruptive is a silicon valley term for technologies like Air BnB which have the potential to upend entire industries, so these might not be the best candidates to work with the NHS. But what is interesting about this list is that most of these apps do not even attempt to interface with the NHS at all, in fact they circumvent it. Doctor Care Anywhere for example might be thought of as an Uber for healthcare, in the sense that users can view doctors, availability and prices on a map on their phone in real time. There is also an app which acts as a social network for in home care. Other apps concern different means of dispensing advice in the absence of a doctor: such as apps relying on genomic data (like 23andMe) patient forums, sleep monitors and prescription drug advice. In other words these apps can be developed independently of the NHS and do not require approval by it to function. But why is this? Is this because the NHS is difficult to sell to or lacks the data infrastructure to implement them?

These websites produced NO networks because, interestingly, the slick, minimal responsive design pages contain few links on them – this reflects a general shift in the way the web works and presents certain problems for the website based analysis of issue crawler: many connections and referrals online are now accomplished through social media channels rather than discreet websites, like the NHS’. This differences between static webpages and more dynamic, interactive sites is sometimes referred to as Web 2.0 versus Web 1.0 though clearly the two logics continue to co-exist.

Billions of people worldwide are suffering from problems for which we have proven behavioural solutions. Yet most can’t access anything other than pills. That’s where we come in. We use tracked data to create highly personalised behavioural medicine programs, all delivered via web and mobile to the highest standards of clinical evidence. From the Big Health Website

What is interesting, in our research so far, is that tech companies like these do not approach innovation in the same way. Companies like Google Deep Mind, who spoke at a recent Royal Society of Medicine event, often start with tools or processes (like machine learning algorithms) and then look for industries to attach themselves to. They, in other words, offer solutions without a problem or solve problems that user’s didn’t know they had.

In this sense they only need to successfully connect with consumers (doctors or patients) rather than embed them selves with particular communities or institutions – or at least this is something which is entailed in what we’re calling the mass personalisation of medicine.

The Local Versus the National

Perhaps a better way to proceed would be to look at one of the self-proclaimed intermediaries between the NHS and the tech community. Ripple is a Leeds-based organization attempting to encourage innovations with data. Ripple also has a helpful page of links called ‘community’. We again gathered the links from this page and ran it through Issue Crawler.

Interestingly, the only links which are shared between this network and the NIC network are NHS.UK, NICE.org.uk and DH.gov.uk (Department of Health). Instead Ripple tends to link to the more local Leeds branches of the NHS and local tech hubs. Also, instead of generalized health and data, Ripple links to several disease specific organisations and charities for depression and alzheimer’s. So another way of conceiving of the innovation problem is as a tension between the local and the national and between specific problems and diseases and more generalised tools.

Now, Leeds is not just any local hub, it is by most accounts, a very successful pioneer community for data based initiatives and the home of the Health and Social Care Information Centre (HSCIC) where centralised health records will some day be stored. However, even if Leeds is a hotbed of innovation, the challenge will be how to roll out or scale up successful initiatives beyond the local, city and trust level – something we’ll be discussing in a forthcoming blog post.

So how do the local hubs at Leeds connect to the more generalised innovation infrastructure of the NIC and do these reveal any unseen bridges between the tech industry and NHS bodies?

Searching for an Underlying Network

Our final network combines the lists above with that of another organisation, King’s Fund, which facilitates medical innovations, particularly of the digital variety. When Issue Crawler crawls these separate lists it may find shared links which do not appear in any of the starting points or a different focal point for multiple networks.

However this crawl did not dig up anything surprising. Again, the research councils dominate this network and even more academic institutions emerge (party due to the influence of the King’s Fund) there are also more condition specific charities and initiativs (Macmillan for cancer) and, interestingly, some links to watchdogs like Ofstead and consumer advice agencies. The HSCIC also emerges for the first time as a central point of connection.

One interesting development though is the size of the nodes for Facebook and Twitter which each have a large amount of total links from the network but not from a variety of sites. This is likely because several websites have Twitter and Facebook buttons embedded at the bottom of every page of their website, not because they are seen as central or authoritative. However, it does raise the point that perhaps the way the tech industry interfaces with the NHS is not through the web, by which we mean static webpages, but through social media or some other (offline) medium or even in back room discussions behind closed doors.

So while this exercise has raised some important questions, perhaps the web is not the the best measure for mapping complex organisations today. The appearance of Twitter and Facebook in the above maps and the lack of tech industry links indicate shifts in how organisations choose to publicise themselves. In a subsequent post we will be looking at how social media has come to structure discussions around digital health specifically in relation to international conferences.

Conclusions

What this cursory exercise has shown is that, at least in terms of web presences, the relationship between the various organisations within the NHS and the tech sector is highly fragmented. There is also a bewildering array of bodies sitting between the NHS and technology companies, evaluation companies, management consultants, funders and tech hubs. There is also a seeming tension between the local and the national scales, where local initiatives are in dialogue with the national level but not necessarily with each other. In other words, this is a very difficult landscape to navigate.

This also highlights tensions between different models of innovation. Is a device and prescription based understanding of technical solutions equipped to handle patient data changes? Is the tech industry’s more personalised model of problem solving missing the existing infrastructure and local knowledge in real clinical settings?

However, we also questioned whether or not the web was the best medium for mapping these tensions. To what extent does innovation happen in public through websites, competitions and public tenders and to what extent does it happen behind closed doors – and in the later case, how are doctors or patients fed into this process? These are the kinds of questions we will be attempting to answer over the course of this seed project.

Does this correspond to your understanding of the role of digital technology within the NHS or not? We’d love to talk to you either way.