Is AI the future of health?

With extreme pressure on resources, the NHS is exploring ways of using its billions of data points to uncover new efficiencies. David Waller investigates how AI is streamlining systems for patients and healthcare professionals through the next generation of chatbots

Innovation

Illustration by Alex Esquerdeiro

Hi, I’m Ada. I can help if you’re feeling unwell.” The sentiment may be familiar and soothing, but this nurse won’t tuck in your sheets or fluff your pillows: Ada is a chatbot, one of a new breed of artificially-intelligent (AI) healthcare apps now tending to sick people across the country.

There’s one other patient for whom this care team may just prove critical: the NHS, which is in less than fine fettle as it approaches its 70th birthday. Its many ailments include long waiting lists, over-stretched staff and nagging demands for £22 billion of efficiency savings from the Government. Proponents of AI argue the technology could make the process of diagnosing and treating problems far more efficient: Ada and other similar chatbot apps act, on one level, as an equivalent to the NHS’ 111 non-emergency number – questioning users about their symptoms via their smartphone, and instantly analysing a mountain of medical data to suggest what might be wrong.

Keeping it personal

“Ada considers thousands of symptoms and conditions, along with all of the user’s information, including medical history and risk factors,” says Bethany Dufresne, a spokesperson for Ada Health. “People are already in the habit of self-informing, with 80% googling their symptoms. But basic internet searches are often misleading and inaccurate. AI has the potential to provide more personalised and accurate advice to patients.”

The rise of such apps – Babylon Health offers a similar service with its GP at Hand app that’s being trialled by NHS Trusts – has led NHS chief executive Simon Stevens to proclaim smartphones “one of the most powerful diagnostic tools available”. The thinking goes that, as people become more familiar with the medical might in their pocket, they’ll also become more aware of their own health and more able to spot and solve potential health problems earlier, which would further help ease the pressure on the NHS.

There‘s also huge potential for Ada’s technology to improve access to personalised health support in regions with less developed healthcare infrastructure, Dufresne points out. An estimated 400m people around the world don’t have access to basic healthcare according to the WHO, but more than half of the world’s population has access to a smartphone. An app like Ada – easy to download and free to use – can help millions of people by making high quality health information accessible to everyone.

A learning curve

Ada’s app was downloaded over one million times in its first six months of being live, according to the company, which raised £36 million from investors. Ada Health now claims that a new assessment is completed on the platform every seven seconds. The program learns from each one, so it becomes increasingly intelligent over time, offering users better assessments and care options as it grows.

Proponents of AI argue that technology could make diagnosing and treating problems far more efficient

This would seem to compare favourably to the average human GP. Cognitive psychologist Arthur Elstein spent a career studying clinical decision-making and found that medical diagnosis is wrong 10-15% of the time, an estimate said by the BMJ to be “very much on target”. This is not to disparage doctors: the total number of known human diseases is in the tens of thousands, so no one could possibly be expected to be on top of every possible combination of symptoms.

With AI on the case, the thinking goes, doctors’ and nurses’ precious time could be freed up to actually care for their patients. “Physicians won’t be replaced,” says Michael Rosenberg, chief commercial and financial officer at Medy Match, a frontline AI application. It helps teams in hospital emergency rooms make better decisions in acute care, analysing raw imaging data along with a host of other information particular to each patient to help doctors make accurate decisions under pressure. “The technology will automate what is a labour-intensive industry, assisting the physician in their daily routine, driving costs down and producing better and healthier outcomes for patients.”

Data hungry

AI and the NHS certainly make good bedfellows – in theory. The technology feeds on data; the NHS has tons of the stuff, gathered through everything from cutting-edge genome research to pensioners’ cataract exams. AI could potentially help the service tackle an epidemic such as obesity, which currently costs the UK £10 billion a year, by analysing all the data we have on the condition and suggesting viable ways to attack it.

Buzz Phrase:

Depersonalised Data

App developers working on research projects with hospitals have to follow strict protocols to protect patients’ identities. The hospital will decide which types of data are made available for research, and all identifiable features, such as patient name and NHS number are removed.

Depersonalised data is vetted by hospital governance teams before it is shared.

Medy Match’s Rosenberg is keen to emphasise that all data is not equal. “We are a big believer in the right data not big data for healthcare. AI can unlock the full potential of patient specific decision support and allow us to discover new relationships as yet undetected by the human eye.”

Untried and untested

But these medical apps shouldn’t necessarily expect the country to simply sit back and say: “Ahh.” The NHS is a beloved institution and the procedure to introduce AI to healthcare has been seen as incredibly high-risk, perhaps even more so than other sectors being disrupted by machine learning, such as finance and the legal sphere. Many believe the money spent on installing an untried new technology across the health service will indeed push humans out of work, and would be better invested in improving conditions so they can deliver a higher quality of primary care.

There are other concerns too. Google’s UK-based AI wing DeepMind has begun partnering with NHS Trusts to trial new medical apps, and activists have expressed misgivings over sacrificing yet more of our most intimate information for the profit of technology giants. Meanwhile, Clinical Commissioning Groups in London recently announced they would abandon plans to triage patients with Babylon’s GP at Hand after they discovered that there was a risk of people ‘gaming the system to achieve a GP appointment’.

The sad reality is that decisions over how to treat a critically ill patient are never less than fraught. The beleaguered, over-burdened NHS is in need of some kind of injection to steer it back to health, and when technology is standing next to the bed brandishing a big needle, it may prove very hard to resist.

Contacting us is easy

Call us on 0345 279 8880

  • Monday to Friday08:30 - 17:30

You will be connected to a member of our team. They will be happy to help answer any questions you may have.

Let us call you back

Please complete as many fields as possible, the fields marked with * are required.