Healthcare circles are buzzing over the development of artificial intelligence – and with good reason. The new generation of AI-powered products and services has the potential to transform not just some, but every aspect of healthcare. At the forefront of medical specialties, radiology is exploring how AI could help interpret images, while consumer-facing apps could become so advanced that they provide affordable and accessible healthcare services to billions of people globally. The smart devices we already use to help control our homes and check the weather could one day detect when we’re getting the flu or feeling sad simply by ‘listening’ to us sneezing or crying.
While this rapidly approaching future bodes well for humanity’s health, it raises questions about trust. Although a symptom-checker app has been shown to outperform some doctors in diagnosing some diseases, can we really trust the app to outperform doctors in every scenario? Another challenge concerns the fast-growing world of digital therapeutics: in the future, the symptom-checker app might not only be able to diagnose but also prescribe another app which we download to treat our symptoms. Software prescribing software, with no human in the loop: what would it take for patients and HCPs to embrace this new paradigm?
We’re creatures of habit and we don’t like change. We don’t like having to think about, yet alone adopt, new habits. We like things the way they are, not the way they could be. Change is difficult and requires time, practice and effort. We are by our nature inherently lazy, and that’s why ease is so important to eliciting behaviour change: the easier, the better. And, ideally, so easy that we don’t even consciously realise something has changed.
Trust in AI varies depending on who you speak to and where in the world they live
Unsurprisingly, some of these new services are being developed by the tech giants. While I trust Facebook when sharing my family photos, do I want it to use AI to screen my posts and determine if I’m at risk of depression, or even suicide? Thankfully, the conversation about what it’s going to take to implement trustworthy AI in healthcare has already begun. I’ve come across two thought-provoking reports: Thinking on its own: AI in the NHS from Reform and Ethical, Social, and Political Challenges of Artificial Intelligence in Health from Future Advocacy.
Trust in AI will also vary depending on who you speak to and where in the world they live. Intel’s 2018 survey found that around 30% of US healthcare leaders stated both patients and clinicians would not trust AI to play an active role in healthcare. Contrast this with PwC’s 2016 survey asking consumers if they would be willing to engage with AI for their healthcare needs: in Nigeria, 94% of people said they would. Poorer nations with fragmented, under-resourced healthcare systems may take the lead in adopting AI in healthcare.
WHO DO WE BLAME WHEN THE MACHINE MAKES A MISTAKE?
The increased use of AI will hopefully reduce the relatively high rate of human medical error that can cause harm, even death. But what if smart machines lead to new types of medical errors? And when they happen, who will be held accountable? If your doctor misdiagnoses you, at least there’s a clear path of accountability. Deep Learning – a branch of AI which aims to replicate the neural networks in the human brain – is already being researched by companies like Google to determine if machines could support decision making in a hospital by predicting what’s likely to happen next to a patient.
Technology is moving faster than our understanding, and if humans can’t interpret how a computer is making a decision, this forms a barrier in its adoption and trustworthiness. Hence the call for explainable AI – AI-powered systems shouldn’t be ‘black boxes.’ From a cultural perspective, it’s fascinating that today we accept the risks in healthcare – i.e. humans making medical errors – but when it comes to AI we’re demanding perfection. A machine has to be completely free of errors and bias before we can trust it. AI is forcing humanity to hold up a mirror to itself and shining a light on our own biases and prejudices.
There’s much debate around ethics and how we build AI that’s free of bias, where existing prejudices around race and gender may filter into the algorithm. While we’re at least becoming aware of the ethical conundrum, a recent study by NC State University concluded that providing a code of ethics to software developers had no impact on their behaviour. So, is the UK government’s code of conduct for data-driven health and care technology going to be sufficient? Are organisations around us failing to keep up with the pace of change? Pfizer has recently appointed a Chief Digital Officer but is that enough in this new era?
Some believe that AI virtual assistants could ‘care’ for people: Amazon Echo, for example, could remind seniors to take their medications as well as help combat loneliness. Children are growing up in an era where it’s normal to have a conversation with an app. I’ve been testing chatbots aimed at children, such as Hugging Face, a ‘virtual friend’ that’s designed to be fun and entertaining, and Replika, which is intended to help young people talk through their problems. The responses, however, aren’t always appropriate. How do we ensure that the virtual assistant doesn’t remind Grandma to take the wrong medication or the AI friend encourage a child to self-harm?
We’re so fixated on regulating the collection of data in healthcare, we forget that consumer-focused AI products will also be collecting health data – with the future direction of healthcare likely being shaped by the organisations that gather the most. Will these organisations really be putting patients first or is it about generating even more economic value? We have to become data literate and understand the implications of how data can be used for collective benefit, as well as to limit our choices.
ARE DIGITAL TOOLS REALLY SHIFTING POWER TO PATIENTS OR IS IT ALL AN ILLUSION?
In 2015 Jeremy Hunt, former UK Health Secretary, talked about making healthcare more human centred: “We have the chance to make NHS patients the most powerful patients in the world – and we should leap at the opportunity.” Certainly, the mantra that frequently accompanies the rollout of new technology is that it’s going to empower patients. Instead of visiting the hospital for a test, we can check for an abnormal heart rhythm in the comfort of our own home. We’ll have an array of apps to detect signs of disease before we even display symptoms. We’ll be able to access our complete medical histories online. This new era of transparency promises to make patient-power something completely normal.
As power moves from providers to patients, will it be evenly distributed?
But what about those patients who find technology a burden, who don’t have the latest iPhone and a data plan? As power moves from providers to patients, will it be evenly distributed? With pressure in every country to keep a lid on healthcare costs, machines that ‘care’ for us could become the new normal. Will the future involve the masses being cared for by machines while only the richest can afford to see a human doctor?
Nations are competing to become leaders in AI, replaying the centuries-old human drama for power and control. However, in this quest to win the race, could nations move so quickly to implement AI in healthcare that corners are cut? Instead of focusing on national AI strategies, perhaps we should be considering how the value from AI can be shared around the world, for the benefit of humanity.
THE CONVERSATION IS JUST STARTING
I appreciate that I’ve posed many questions without providing many answers, but there are so many unknowns at present. Society’s response to developments in technology and data isn’t as rapid as it could be. To build trust, we must first understand why people are afraid of AI. Then, instead of dismissing them as luddites, we must genuinely listen to their concerns and invite them to be part of the way forward.
Given that AI will be impinging on so many aspects of healthcare, there are complex challenges relating to accountability; our laws, for example, will need to accommodate a world where machines tell doctors (and patients) which treatment will be most effective. With the current concentration of power in healthcare, AI could become the catalyst that accelerates the shift to a future where patients have genuine power, both individually and collectively.
At this point in the 21st century I believe we stand at a crossroads. While, like many, I’m impatient for change to happen, I’m also mindful that this can’t be led by the privileged few. There’s a wonderful opportunity to involve all parts of society in building a future where we can harness AI in a trustworthy manner – providing healthcare that’s adaptable, accessible and affordable to everyone on the planet.