Report identifies barriers to using artificial intelligence in healthcare

22 Jan 2018 | News

Despite the hype, there are few examples of the successful application of AI in health services. A different approach to adoption is needed, with access to data for companies and a framework to ensure proper testing and monitoring of AI systems

Robots won't replace your doctor yet. 

A new report illustrates areas where artificial intelligence (AI) could help the National Health Service (NHS) in the UK become more efficient and deliver better outcomes for patients – highlighting the main barriers to the implementation and suggesting some potential solutions.

According to the report, by researchers at the think tank Reform UK, despite the hype around AI in healthcare, examples of it being implemented and deployed in the NHS are sparse.

The problem is that as things stand, it is incumbent on individual healthcare providers to introduce new technologies. This has resulted in piecemeal applications and patchy realisation of benefits.

With a different approach to technological adoption, in which AI is gradually embedded service transformation plans that are currently in the works, the future could look quite different, the report says.

AI could support the delivery of the NHS’s Five Year Forward View, which aims to narrow three gaps in health provision:

  • AI could help address the health and wellbeing gap by predicting which individuals or groups of individuals are at risk of illness and allow the NHS to target treatment more effectively towards them;
  • The reduction of the care and quality gap could be supported by using AI tools to give all health professionals and patients access to cutting edge diagnostics and treatment tailored to individual need;
  • AI could help address the efficiency and funding gap by automating tasks, triaging patients to the most appropriate services and allowing them to selfcare

Improving buy-in for artificial intelligence

In order to apply AI to support a more efficient healthcare system that delivers better outcomes, it is first necessary to overcome concerns of both the public and healthcare professionals.

Public confidence and trust are vital for the successful development of AI, says the report. Building this trust must start by increasing public confidence in the way data is shared both within the NHS and with external organisations.

Data quality is another critical requirement. The NHS needs to get data right to truly harness the potential of AI in healthcare. This means collecting the right type of data in the right format, increasing its quality and securely granting access to it.

As things stand, the healthcare system is still heavily reliant on paper and most of its IT systems are not based on open standards. This is not only limiting the exchange of information across the system, but also holding back AI development, because the accuracy and fairness of AI algorithms are wholly dependent on the data the algorithms are fed.

Ethical and safe use are also critical to the deployment of AI, and should be a central matter of interest for regulators, including the National Institute for Health and Care Excellence (NICE), the Medicines and Healthcare products Regulatory Agency and the government.

Healthcare is a high-risk area, where the impact of a mistake could have profound consequences on a person’s life. AI systems are not infallible or devoid of bias, the report notes. It is important for current regulations to be updated to make sure that the applications of AI in healthcare lead to a better and more efficient health service, which reduces variations in the quality of care and in health outcomes.

How to make AI deliver for health

The report recommends that NHS Digital, as the national provider of IT systems, and the 44 Sustainability and Transformation Partnerships (five-year plans covering all aspects of NHS spending in 44 defined geographic areas in England) should consider producing reviews outlining how AI could be appropriately and gradually integrated to deliver service transformation and better outcomes for patients at a local level.

But, says the report, AI is an enabler, not an end in itself, and caution should be taken when embedding AI within the service transformation plans. It should not be regarded as tool that will decide what objectives or outcomes should be reached.

The NHS England and NICE should set out a clear framework for the procurement of AI systems, to ensure they are not so complex to use that they actually hamper service transformation and become burdensome for healthcare professionals.

To support this, the report says NHS England and NICE should consider including the user-friendliness of IT systems in the procurement process, and favour intelligent systems that flag-up errors in real-time.

To smooth the way for implementing AI, all healthcare IT suppliers should be required to build in interoperability from the start, allowing healthcare professionals to migrate data from one system to another.

The report also suggests that NHS Digital should commission a review to evaluate how data from technologies and devices outside the health and care system, such as wearables and sensors, could be integrated and used within the NHS.

At the same time, NHS Digital should create a directory of training datasets, such as clinical imaging datasets, which could be made available to companies to train their AI algorithms. NHS Digital needs to specify the conditions for accessing this data.

In addition, there needs to be national framework setting out the conditions under which companies are able to derive commercial value from patient data, to ensure this happens in a way that is beneficial to the NHS.

Never miss an update from Science|Business:   Newsletter sign-up