This blogpost draws upon fieldnotes gathered at the 2016 Recent Developments in Digital Health 2016 conference hosted by the Royal Society of Medicine (RSM) to explore some of the tensions that arise between ways of talking about medicine and about tech startups.  

Digital data and healthcare make strange bedfellows. Watching tech entrepreneurs and doctors come together is reminiscent of ‘bundling’, an almost forgotten courting ritual in which near strangers were, sometimes reluctantly, wrapped up in bed together by their seniors (we didn’t make this up). The RSM’s Recent Developments in Digital Health this past February was one of many events which put these strangers together in one room. And while the tone was generally upbeat, bordering on tech-utopian at times, it also bore many of the hallmark tensions of different worlds colliding.

 

RSM Cola

The collision begins with the RSM building at 1 Wimpole Street itself. The Society has been there since 1909 but since 2014 the building has shared a footprint with Coca-Cola’s striking new EMEA headquarters. Given the potential role of sugary drinks in the global crises in preventable disease (for example here and here) some visiting delegates were initially disoriented by this very close, incongruent proximity. First impressions aside, the two have got things in common. Coca-Cola’s origins as a proprietary health tonic, a ‘patent medicine’, is well known. Neither party shouts about it these days but before its late nineteenth century professionalisation, medicine was a mixed bag of science, quackery, lay knowledge and market speculation. History has some salutary lessons for anyone eager to purify health and healthcare from the taint of filthy market lucre. Healthcare always has a business end. Even when the NHS was much closer to Bevanite founding operational principles than it is now, there were encounters with markets.

Cola

But the digital health world has truly upped the ante. You don’t need the painful lessons of the NPfIT to concede that the scope and scale of innovations being talked about could not reasonably be accomplished within any single healthcare system. Digital health is challenging, exciting -and really hard to get working at scale- because it is happening in the weird alchemy of the encounter between provisionally separated worlds. The political work of decades that went into clearing space between publicly funded healthcare on the one hand, and the profit-seeking industries and technologies surrounding wellbeing, leisure, sports, entertainment, motoring, food and all manner of bodily comforts on the other, is now coming under an accelerated threat. The progress of this threat is however far more complicated than shorthand terms like ‘marketization’ and ‘privatisation’ can capture.

We were packed into a wood panelled oblong room with opulently padded leather-ish chairs that shamed their occupants by squeaking noisily on the slightest movement. The sold-out event was organised by the ‘Telemedicine Group’, which just shows how new innovations can take root in the space claimed by past innovations. The speakers included government leaders, NHS managers, clinicians, academics, patients and start-up executives and the audience was just as diverse.

The opening speaker, Tony Young, straddles many of these roles. Young developed several commercial devices in his medical practice and is now the NHS National Clinical Director of Innovation. Dr Young spoke with all the nonchalant hubris of an entrepreneur in the Elon Musk-mould, claiming, with a cheerful smirk, that his aim was to improve the lives of everyone on the planet. He explained that one thing digital health can provide is a path toward preventative care. The NHS currently focuses mainly on acute care but with a growing, ageing, sickening population, the biggest coming challenge in England, as elsewhere, will be tackling chronic, preventable illness.

Young talked also about NHS funding challenges and offered an extensive list of companies developing digital health products. These ranged all the way from sterilised iPad cases to apps. How these disparate innovations might connect together to improve lives was not explained. Equally unclear was how such innovations could be distributed across the highly fragmented healthcare archipelago of NHS trusts, CCGs, local authorities, voluntary and social care organisations etc. This is the infrastructural work of public administrative science, procurement, reimbursement, informatics, coding, standards, workflow analysis etc. It is not clickbait and it is not being talked about in keynote lectures. But if digital innovation is to be concretely related to the funding challenge this has to be worked through somewhere. (If this work is being done systematically, at policy defining levels anywhere in NHS England’s thousands of parts, we have yet to find it. We’d really like to, so if you know something we don’t – please get in touch ! [payingforhealth@gmail.com])

Impressed, the moderator of Young’s session, exclaimed ‘maybe we should rebrand the NHS as the “Novelty Health Service”!’ He was playing to the audience – though it was not clear which of the many audiences assembled he had in mind. One of the first questions from the floor, starting a pattern that repeated throughout the day, came from the CEO of a health tech start-up, who used the space to pitch his company. Other questions followed from junior doctors who talked passionately about the need for digital training and information in medical education.

Martin Kelley, who works for tech incubation hub Digital Catapult, one of many facilitators in what we call the ‘liminal spaces’ of digital health innovation, spoke about the challenges of developing new products. He gave the example of a ‘hack day’ with end users, clinicians and app developers in San Francisco around the topic of dementia. Tim Ellis of the Department of Health spoke next about the centralised monitoring of innovations in the NHS. The Digital Maturity Self Assessment Tool was designed to give benchmarks for local trusts to determine if they are keeping up with best practice. At this point, a stately tweed-clad doctor in the back asked, to no one in particular, why digital was assumed to be better in every case. He cautioned that doctors hunched, typing over desktops were losing touch with the caring arts of the bedside manner. If the EHR backlash in the US is anything to go by he has a point. Though, as a tweet by Francis White, another speaker and Vice-President at AliveCor, hinted later that day – paper too can detract from patient face-time. Exasperated doctors took several opportunities to take potshots at their governmental managers throughout the day.

 

 

The story of the NHS is one of complex negotiations between the state and the medical profession. In its post-war formation, resistance to the concept of a nationalised service came, not from rival political parties, but often from the medical establishment. Ultimately, medical groups, had to be worn down and bought out, just as they had in the earlier struggle to enact the 1911 National Insurance Act. In much the same way Matt Taibi argued the insurance industry was bought into Obamacare with promises of guaranteed customers.

The distribution of responsibility for healthcare payment across individuals, markets and governments courts an always simmering controversy on both sides of the Atlantic. ‘Too much market’ controversies in the UK are echoed by ‘too much government’ controversies in the US and the balance between public and private subsidy in both places is publicly misunderstood.

 

 

After the break, the room hushed as Mustafa Suleyman from Google DeepMind took the stage. The moderator reminded us that there would be no photos during this presentation. Ironic given Google’s hunger for every morsel of everyone else’s data. Suleyman started by explaining how DeepMind’s hierarchical neural networks learned to play the entire battery of Atari arcade games – without any human advice. Machine learning moves beyond the brute force approach, which allowed Deep Blue to beat a chess grandmaster, to a more dynamic and adaptive process. This process was just about to be tested on a Go grandmaster, a match which DeepMind subsequently won, but some think the NHS may be a more formidable challenge.

Suleyman talked next about working with doctors in the ‘co-design’ of patient monitoring apps that collate all the relevant data in a usable interface. Most medical data is ‘pull’ – the information must be requested, when it should be ‘push’ like phone notifications. User experience design has progressed to a point that apps are expected to be intuitive and simple. As impressive as the presentation was, the relationship between machine learning and a phone app were not exactly explicit. Uses are not yet clear though and many remain contentious. Is AI about creating digital algorithmic triage or about auditing behaviours? If companies have proprietary rights to diagnostic algorithm who gets access to what’s in the black box?

And who has access to the data? If advocates are even half right that #datasaveslives, these deliberations do need to happen. They have to happen in the right places with the right people paying attention. But how exactly? There is already no shortage of debate forums nor of data sharing expertise.

Google’s presentation featured many of the typical assumptions surrounding ‘big data’. Of course ‘unsupervised’ machine learning is good if you’re trying to develop AI – but does good equate to more objective? Why not make use of the tacit knowledge and expertise that doctors already possess? Similar rhetoric was used in the presentation by Euan Ashley, a cardiovascular specialist involved in the Apple HealthKit based MyHeart Counts study. MyHeart Counts asked users to perform simple tasks, such as walking a certain number of paces, with their phones and the resulting readings became one of the biggest samples in the field. Some in the audience raised concerns about data quality and the skewed sampling of iphone users only (in the first study). Ashley’s defence, that the sheer amount of data corrects other bias, is becoming common in big health data but some insist that scaling up bad assumptions only magnifies them. But studies like MyHeart Counts are defining new normative practices that challenge the established gold standard of Randomised Controlled Trials. MyHeartCounts was also interesting because it pioneered variable consent – where users can tick boxes to give some, but not all, of their information away. Later in the day, legal scholar Susan Wallace observed that the most pertinent law governing personal data was written in 1998.

In the afternoon we got a user-focused argument from Roz Davies, representing the We Love Life social enterprise. She made the case that among the biggest benefits of digital technology is empowering patients. Dr Ameet Bakhai then talked passionately about what gets left out of digital health initiatives: the seven C’s (confidence, confidentiality, candour, courage, collegiality, compassion, complications – some patient groups might add culpability).

7Cs

The final session on sensing technology went back to business. AliveCor’s Francis White, flashed an intentionally provocative slide (above) on some of the challenges faced at the business end of the health sector. AliveCor’s mobile EKG device is among the most clinically validated of recent health tech innovations and detects a condition, atrial fibrillation, that is common, dangerous and frequently missed. The obstacles to scaled adoption within primary care of devices like AliveCor’s are incentivising health care pathways that bypass the NHS. Their device is available in certain NHS pilot schemes, but more of their stock is shifted in the UK through Amazon than through the NHS.

This fascinating event was emblematic of the trouble brewing in the NHS/market/tech encounter. If the hype is ever to match the circumstances of healthcare delivery its not innovation that’s the problem its defining the rules, standards, ethics of the engagement.