‘Data saves lives,’ says Maxine Mackintosh. ‘It may not be as direct or as dramatic as the operating table, or in A&E... But when the next Thalidomide happens - how long is it going to take us to work out where the problem lies?’
‘If health data is liberated and can be analysed by the best minds and machines’, she explains, ‘warning signs could be spotted before they become full blown crises’.
‘Your data really can save your friend, family or neighbours’ lives and that needs communicating. Articulating such success stories is hugely important in demonstrating the benefits and winning public trust around the effective analysis of health data,’ she asserts.
Working two jobs
Mackintosh, by day, works on her PhD in dementia and data science at the Farr Institute of Health Informatics, UCL. By night she’s the Chair of HealthTech Women UK - the British branch of an international professional network that supports and promotes women to be the future leaders in health innovation.
Passionate and palpably excited about data’s ability to revolutionise healthcare, Mackintosh paints a compelling picture of tomorrow’s world. It’s not a sales pitch though - she’s equally forthright about the challenges the sector faces as it makes its essential march towards digitisation.
‘We’re nearing the peak of the hype cycle - digital health is still quite flashy and a lot of it is the digitisation of the status quo,’ she says. ‘I’d like us to quickly move into transformative technologies that really save lives, that saves money and keep healthy people healthy.’
For Mackintosh, this means creating a healthcare system that is designed to keep people healthy rather than make them better. ‘At the moment,’ she says, ‘we have a sick care system that’s reactive - you interact with it when you’re ill. The esoteric ideal is that we’ll live in a purely predictive and preventative environment. It means that you keep healthy people well.’ And at the heart of this great transformation is the easy passage of data.
Rough roads behind and ahead
Moving quickly towards this future isn’t going to be easy. One reason is the public is understandably mistrustful of sharing its data. ‘Healthcare data is vulnerable,’ she says. ‘It’s the most sensitive data we have.’ Set this fact against a backdrop where news headlines shout about hacks, losses, dubious data trading practices and it’s easy to understand why people don’t want their personal data shared.
Care.data is a prime example of, on one hand, the need to enable health data to be shared and analysed for population health management. And, on the other, the public’s reluctance to let their information be pooled and probed. Initiated in 2014, care.data was an ambitious plan to create a vast database that would hold the health records of everyone in Britain.
The need to communicate
Despite its lofty ambitions, care.data was doomed from the start. Rather than nurturing public approval through carefully communicating the project’s possibilities, Care.data was flawed from the onset by a poor PR campaign. Huge swathes of the population remained unaware of care.data’s existence. The problem was compounded by an information leafleting campaign that fell well short of reaching every household.
These communication failings meant the first thing many people heard about the project was through blowhard headlines about the risks of hacks, data losses and allegations of Machiavellian manoeuvrings by insurance companies.
‘Given the headlines, we may see a parallel with DeepMind now,’ she says. ‘And like Care.data, a lot of the problem lies in poor public understanding of how the NHS is structured, versus necessarily how their data is used.
DeepMind is one of hundreds of private organisations operating in the data and supplier space. No one batted an eyelid before. But as we now have a captive audience, let’s get more people talking about our health data’
DeepMind is Alphabet’s artificial intelligence and research platform. It’s designed to make scientific breakthroughs in complex spheres such as healthcare. Just like Care.data, it requires huge pools of information.
The benefits need to be explained
‘People are irrationally nervous,’ she says. ‘They’re not comfortable with the concept of companies accessing their health data. Yet less than 20 per cent of the public are aware it’s happening in the first place. Basic public understanding and trust needs to be built.’
The fact remains though that today over 90 per cent of patient data in the NHS is dealt with and analysed by private organisations. This gut reaction to big business managing big data is wrong, Mackintosh asserts. ‘We need,’ she says, ‘the best statisticians, data scientists, cybersecurity specialists, ethics and information governance experts all around the table.
At the moment, a lot of that talent is being attracted to the Googles and Facebooks of this world. We can’t ignore that that is where enormous expertise lie.’
Just as importantly she says: ‘We need to work really hard to articulate what can be achieved and what the opportunity cost of not sharing and probing our health data is.’
The value of big data in health
So, what, in concrete terms can big data hope to achieve in health? Again, Mackintosh is realistic.
‘Big Data is a really hot topic across many sectors - it’s going to solve everything, so we’re told.’ Despite the hype, she says: ‘I think there’s, of course, huge potential in the breadth, volume and pace at which data is being collected in health.’
There are, she explains, three big benefits for health. Firstly, there’s the potential sources of data - everything from medical records, to wearables, to your internet activity data. ‘We can draw all those sources together and start to paint a really accurate picture of people’s health, before they get sick. It creates a much fuller picture for disease prevention.’
‘Secondly’, she explains, ‘data will let us rethink systems. You can understand a story by looking at how data flows, by looking at what is happening across a health system. That lets us, for example, allocate resources better and see where burdens and inefficiencies lie’
Finally, she explains that big data lets us ask different questions. Previously health researchers carried out traditional epidemiological studies - work that involves the analysis of patterns, causes and effects.
‘Now,’ Mackintosh says, ‘we can use novel approaches, including machine learning. We can use hypothesis-free methods and let the data speak for itself’
Making it all happen
Along with ensuring the public is comfortable with the idea of its data being shared, Mackintosh believes the health sector itself will need to make adaptions and allowances.
‘What’s exciting about the digital health space,’ she says, ‘is it looks to transcend traditional silos such as academia, business, the health system itself, and the start-up world. You can find novel solutions in the gaps between sectors.’
Mackintosh believes that there’s much to be learned from Silicon Valley. ‘I think we’re quite complacent in the UK about our NHS - we don’t always question it, how it works and how it functions. I think we should be looking to engage with the public about these issues.’
Mackintosh isn’t, however, calling for a swashbuckling approach to organising digital medicine. Rather she recommends caution. ‘There are two views on professionalising a new industry,’ she says. ‘On one hand you might be restrictive, you might prevent innovators from other industries coming in.
On the other hand, we can’t ignore the fact that digital health sits within healthcare which is rightly, a very highly regulated market. So, it’s about finding the right balance - being an open profession that welcomes people from different sectors and technology backgrounds.’
Critically, she says, we must professionalise digital health in a way that lets us adhere to healthcare’s principles.
Data and dementia
‘I spend my days doing my PhD at the Farr Institute,’ Maxine Mackintosh says. ‘I’m working on mining medical records to find new predictors for dementia.’
A key problem with dementia is that the disease causes a slow and progressive decline in cognitive abilities, she says. The point at which you’re diagnosed is quite arbitrary too. Usually though it’s at the point where you have significant memory impairment and, by that point, it’s too late to intervene.
‘The problem is we don’t know what early dementia looks like,’ she says. ‘And I’m interested in understanding the factors that are indicative of cognitive decline… Is it that early dementia patients stop attending appointments? Do they see their GP more? Or, do they have severe bouts of anxiety, depression or cardiovascular concerns? Are they hospitalised because of accidents or falls? And what about the combination of all of these?
It’s that huge breadth of healthcare interactions which, when you put them all together, can create an accurate cognitive imprint of what that pre-clinical dementia period looks like. And, we’re able to do that because of the data that’s routinely collected in medical records. We’re uniquely able to do this in the UK because we have some reasonable quality, longitudinally collected medical data.’
