• Nesta warns of danger that, without public involvement, AI technology could make NHS harder to access, or squeeze out doctor-patient dialogue
• New survey shows that 85% of the public have already consulted Google about symptoms when unwell, and 1 in 3 say they would be comfortable having a conversation with an AI app about whether they should see a doctor
• Urgent NHS action needed to use ‘People Powered AI Principles’ and involve patients ‘before it is too late’

2 May 2018: Apps which use Artificial Intelligence for diagnosis and triage could be commonplace in GP surgeries and A&E within five years, if artificial technology fulfils technologists’ predictions, according to a new report released today by Nesta.

However, Nesta’s Confronting Dr Robot report warns that, whilst effective AI advice and triage could be very appealing to both a health service that is under severe pressure, and to patients that are already routinely googling their symptoms, there are big risks in putting machines between patients and doctors.

As AI looks set to become the front door to the NHS, it has to be developed in a way that will help and not hinder healthcare. Nesta urges NHS, government and doctors to ‘act before it’s too late’ by involving patients and practitioners in the design of this rapidly-emerging field of automation and artificial intelligence (AI) in healthcare.

The report follows last week’s appointment by Jeremy Hunt, Secretary of State for Health and Social Care, of Dr Eric Topol - an expert in cardiology, genetics and digital medicine - to look into how tens of thousands of NHS staff could be trained to use artificial intelligence and robotics. *

Confronting Dr Robot says that the decisions being made now will determine if AI technology could either:

• free up capacity, reduce waiting times and make it simpler for patients to get to see a doctor
• Or, risk making the NHS harder to access, squeeze out doctor-patient dialogue and adversely affect people with complex social and health needs.

Pressure is mounting on the NHS to find solutions to capacity problems, particularly in GP and A&E services. Around one in five GP and A&E visits* are for minor ailments or conditions that could be treated at home (sprains, ‘flu and insect bites are some of the largest categories of self treatable conditions that come to A+E). If AI is to alleviate pressures through a triage role - and if patients are to be confident in accessing it - the technology and the medical advice needs to be robust.

The report is published as a new national survey by Populus for Nesta shows:

• 85 per cent of people say they have consulted Google about their symptoms when feeling unwell, and of these, 68 per cent say they have done so in order to decide whether they should see a doctor
• one in three people (34 per cent) said that they would be comfortable following suggestions by a health app on things they could do to help manage the health issue themselves, if the app concluded they didn't need to see a doctor
• Yet, 44 per cent of people say they would be uncomfortable having a conversation with an artificial intelligence app or website about whether or not they needed to see a doctor, and of these, 65 per cent said that they would be concerned that AI wouldn’t understand their needs

While services struggle with capacity issues, tech giants are seeking to develop potentially huge tech-health markets, and are scrambling to harness big data and AI to develop systems to address the demand.

Report author and head of strategy for Nesta’s Health Lab, John Loder, underlines the need for NHS and government to act now to ensure patients are at the heart of technology firms’ development of services: “Within five years, AI could be the first thing many people using the NHS might encounter - and it has the potential to make things simpler, more accessible more responsive and put patients in control. But, if patients and doctors are not put in the ‘driving seat’ of its development, and soon, the technology risks creating more of a barrier than an open door, offering opaque advice and dehumanising healthcare.”

Geoff Huggins, director for digital health and social care, Scottish Government, adds, "It's clear that there's great potential and opportunity for technology to better assist practitioners and patients alike. But we have to do this in a way that is complementary and not simply introduce technology because it exists and is possible.Striking the balance where technology gives people more power and better results will be the key."

The Nesta report highlights three main areas where AI is likely to be applied by the NHS, and the consequences for patients, depending on how well it is developed and implemented:

• Direct to consumer advice and triage. AI could enable an app or a ‘chatbot’ to elicit symptoms and interpret these to offer self-management hints or advice to seek further medical help. This could help solve a major issue of inappropriate use of limited healthcare resources. However, it could also generate a flood of unnecessary demand from false positives or risk-adverse advice, and could widen health inequalities - for example by disempowering those without access to technology.
• Giving early signs that a health problem is worsening. AI can extract complex data in real time and inform patients that they need to seek urgent care - for example, by assessing breathing sounds of people with congestive heart failure to spot signs of deterioration. This could make the health system more dynamic and responsive. However, it might replace individualised conversations or generate unnecessary concerns.
• Getting an automated second opinion. AI could sit alongside doctors to offer a second opinion on diagnosis and treatment of a patient, or enable patients themselves to seek a second opinion. This could empower patients and improve healthcare decisions but might also change the power dynamics in health, and create additional cost pressures.

Loder concludes, “The danger is that the NHS reaches for what it’s offered by the commercial market to solve some its problems - but that could lead to creating other problems elsewhere if systems are not designed around the needs of patients and front line staff. What Nesta wants to see is a people-powered and data driven health system that makes the most of resources - from patients’ motivation to data, as well as the knowledge and commitment of health professionals ”

The full report is available to download at: www.nesta.org.uk/publications/confronting-dr-robot

- Ends -

For further information, please contact:

Nigel Campbell, Media Manager, Nesta on:

07825 4452339 / [email protected]

Notes to editors

• Populus interviewed a Nationally Representative sample of 2,057 GB adults aged 18+ from its online panel on the 24th April 2018. Populus is a member of the British Polling Council and abides by its rules. For more details go to www.populus.co.uk
• *Research published in 2010 showed that 1 in 5 (20%) GP appointments - 51.4 million consultations - were solely for minor ailments which could be treated at home (including dandruff, heartburn and fungal nail infections), and this was true for almost the same proportion (19%) of A&E appointments, according to research published by the Proprietary Association of Great Britain, the UK trade association representing manufacturers of branded over-the-counter medicines, in 2015.
  
  • http://selfcarejournal.com/wp-content/uploads/2015/09/IMS-1.3.105-16.pdf
  • https://www.pagb.co.uk/content/uploads/2016/06/PAGB_AE_Exective_Summary_June-2015.pdf
• * 27th April 2018. A review has been announced into training NHS staff to use AI and robotics. The plans are part of an independent review into the training needs of NHS staff to ensure they are equipped to give patients the latest treatments: https://www.gov.uk/government/news/review-announced-into-training-nhs-staff-to-use-ai-and-robotics
• Artificial Intelligence systems are already being trialled for introduction into some areas of the NHS . In 2017, ‘GP on Hand’ an AI-based system was trialled in London as a partial replacement for the NHS 111 phone triage system, in which 1.2 million people were offered the option of typing their symptoms into an app which would judge whether they should see a doctor.
• The Nesta report makes four key recommendations to ensure that ‘People Powered AI’ principles are used as AI technology and systems are developed for the NHS:
  
  • 1/ Put citizens and health professionals at the heart of the design, development and implementation of any system. This should include public panels to ensure the technology caters to the needs and concerns of citizens, and that health professionals are widely consulted
  • 2/ Test in real-world, complex conditions which accurately recreate actual uses and compared with non-AI approaches, and to learn and adapt before wider take-up
  • 3/ Implement regulation that anticipates market conditions, such as who owns both algorithms and data, to mandate adherence to key system design principles and ensure that a diversity of entrants come into the market to avoid a monopoly from a few tech firms
  • 4/ Ensure that decision-makers in NHS, government and at service-commissioning level have the technical skills and power to scrutinise, manage and deploy AI in a responsible way. This could include inclusion in medical education, health management training and front line staff professional development/training.

About Nesta

Nesta is a global innovation foundation. We back new ideas to tackle the big challenges of our time. We use knowledge, networks, funding and skills - working in partnership with other, including governments, businesses and charities. We are a UK charity but work all over the world, supported by a financial endowment. The kinds of big challenges we tackle include ageing, stretched public services; a fast changing jobs market and people feeling disempowered. We focus on areas where the combination of digital technology, empowered individuals and better use of data and evidence can have the biggest impact.