The coronavirus pandemic has caused significant shifts in how healthcare is delivered. How many of these advances will last beyond the virus?
The coronavirus epidemic has put healthcare around the world under tremendous pressure. It has also forced new ways of thinking and working on to organizations that have traditionally moved very deliberately and cautiously when considering innovation.
COVID-19 has forced hospitals, GPs, and social care organizations to consider everything from video consultations to robotics to help keep services operating. But how significant will these changes be in the long term? ZDNet looks at the shifts that the coronavirus outbreak has caused, and how these will likely change the future of healthcare in new and unexpected ways.
More AI, more machine learning
Artificial intelligence and machine learning have been deployed in various ways to try to lessen the impact of coronavirus: from finding drugs that may be able to treat the disease to identify early signs of the condition on medical images.
Work is also ongoing to use AI as a tool to help clinicians with decision-making: identifying which patients may be safe to discharge, or which are likely to deteriorate and need hospital admission, for example. The University of Maastricht, New York University, and the University of Chicago are among those developing AI that can be used to triage COVID-19 victims.
However, Eric Horvitz, chief scientist of Microsoft, believes that in future AI could also be used to help with patients’ decision-making too, by helping develop more accurate models of risk. “This is an opportunity for harnessing machine learning to help people across the world answer questions like ‘what is the risk [to me] if I become COVID-positive and what is the risk if I engage in a specific activity?'”
“Predictions of risk will be valuable in public discussions as well as in designing policies,” Horvitz told the recent COVID and AI The Road Ahead event at Stanford University’s Institute for Human-Centered Artificial Intelligence (HAI).
By setting AI to work on developing risk models, public authorities can create predictive systems that can determine which mix of lockdowns, hygiene measures, social distancing and shielding, addressing healthcare disparities, and test-and-trace systems will be most effective in dampening coronavirus outbreaks within particular communities.
Once such systems have been set up and rolled out for one condition, the work involved in repurposing them for different diseases – either the next pandemic or more common medical conditions – is relatively small. That means that once AI systems have found their place in healthcare institutions due to coronavirus, there use is likely to grow significantly from there.
Of course, AI systems are only as good as the data they’re trained on, and if there is one thing that researchers have discovered during the COVID machine learning push, it is that health data – even in a global pandemic – is all over the place, buried in proprietary, messy, and non-interoperable silos.
Susan Athey, the economics of technology professor at the Stanford Graduate School of Business, told HAI that research she has been working on – to determine whether the medicines a person was taking before they caught COVID have a bearing on how the disease affects them – was hamstrung by the fragmented nature of health data in the US.
“The data problems were just enormous,” she said, due to the disparate and inaccessible sources of relevant information.
“Without the data, you can’t do any machine learning, let alone high-quality machine learning… it is a huge policy issue that we need to solve now, but also for the future,” Athey added.
There are signs that the regulators are waking up to the challenge. In the UK, for example, the health service’s digital arm NHSX recently published the COPI notice, which streamlines how organizations can use public data in the fight against COVID-19. The changes in data use have already aided the work of the OpenSafely project, an open-source analytics project that allows researchers to query 24 million NHS patients’ records.
Many where else, tech companies and health organizations have also worked to publish coronavirus-related datasets for AI and machine learning, including, most notably, the COVID-19 Open Research Dataset.
Telehealth and remote medicine
Calls for greater use of telehealth, where patients can see their doctor remotely via video conferencing or similar, have been around ever since home broadband became a reality, and the response to the coronavirus has made this much more likely.
While telehealth tools are being rapidly rolled out, their effectiveness depends on whether doctors are willing to change their processes, said Beccy Baird, a senior policy fellow at the King’s Fund.
“If you are just adding it in, it creates more work, but if you totally redesign the process, it can be incredibly helpful,” she said.
“Having the headspace to do it is very important, it is very hard to do this kind of massive change on top of the day job when you feel overloaded already and we have a workforce crisis in general practice,” Baird added. According to NHS Digital, coronavirus led to the planned rollout of online consultations happening at a faster pace: before the virus, around 1,200 GP practices were expected to have the system in place by the end of June; instead, it was rolled out to 2,200 practices by the end of April.
In the US, meanwhile, the use of telehealth is also expected to rise 63% this year due to the coronavirus outbreak, according to analyst Frost and Sullivan.
A move to DIY and open source
As the coronavirus pandemic threatened to overwhelm health services earlier this year, a number of grassroots initiatives sprang up to try and plug holes in health supply chains. Local maker spaces began to 3D print in-demand face shields for front-line workers, while healthcare start-ups and staff set to work creating open-source designs for ventilators as a shortage of the vital equipment loomed.
According to James Barlow, professor of technology and innovation management (healthcare) at Imperial College Business School, such homegrown innovations can struggle when they begin expanding. “The problem with a lot of them is that they fail at the scaling-up stage. It is a good idea, they may get a little bit of funding to support the development work and the initial pilots, but then they fail because they can’t raise sufficient funding to do trials at a large scale and gather robust evidence for the benefits,” he said. However, Barlow notes that between one-quarter and one-third of medical innovations are driven by either patients or front-line medical professionals. “As an approach, it is certainly here to stay I think,” he added.
The increasing use of low-cost, streamlined, and homegrown tech also builds on the trend for “frugal innovation”, which predates the pandemic.
Over the last five years in developing economies, there has been consistent innovation in “very simple solutions that work in the context of the region low cost, decent technology that does the job,” Siddharth Shah, program manager at analyst Frost and Sullivan, said.
“It did not have to be a billion-dollar developmental process, a million-dollar cost associated with acquiring that concept. And frugal innovation had become big. I think this pandemic will prove the case that everything does not have to be big and fancy and you do not need to spend millions and millions of dollars developing it. Overall, I think that move will get accelerated,” he added.
With DIY and homegrown projects often using open-source approaches to allow other groups to reuse their work, the idea of co-creation appears to be rubbing off on commercial entities too.
“There may be a new trend developing that goes beyond just making IP available, and it extends to enabling others to use and fully exploit the IP,” Noel Courage, a partner at IP law firm Bereskin & Parr, wrote. “It is a big deal,” he added.
Elsewhere, the Open Covid Pledge initiative, which is encouraging companies to share IP that could be used to fight COVID-19 for free, has drawn support from the likes of Amazon, Facebook, Intel, IBM, and Microsoft.
To commercial organizations, IP is viewed as somewhere between their crown jewels and state secrets. However, that doesn’t mean that the idea of sharing IP might not be here to stay, at least in the medium term.
According to Frost and Sullivan’s Shah, the many post-coronavirus financial struggles of healthcare providers will encourage the model to persist.
“Generally, there is probably going to be a move toward that kind of a situation that open innovation, sharing of intellectual property – in most cases. This is not because people saw the better side of humanity, it is also because there is going to be economic pressure this pandemic is putting so much pressure on healthcare systems. For a couple of years, at least, healthcare systems are going to be negative in terms of their budgets and financial situation,” Shah said.
Source: ZDNetRelated posts: