Can AI help put people firmly at the centre of their care?

The NHS.uk website averaged over 2,000 visitors per minute in 2022 and, while websites are hardly considered cutting edge, this technology is important to help make trusted and reliable health and care knowledge easily accessible to patients and the public. Web-based information, alongside access to medical records and personalised care initiatives, means people are potentially more informed to make decisions and be actively involved in their own care.

'Up to 61 per cent of adults have low health literacy levels which means they may not be able to understand or use the information available to them.'

However simply having access to information doesn’t necessarily make it useable. Health literacy refers to the skills, knowledge and confidence to understand and use health care information. Up to 61 per cent of adults have low health literacy levels which means they may not be able to understand or use the information available to them. This can impact people’s capability to have shared responsibility for their health, resulting in a lack of independence and autonomy and an overdependency on services. It’s difficult to measure the financial impact of low health literacy on a health care system, but some estimates put it at 3–5 per cent of health budgets. With the well-established increase in long-term health conditions and an ageing population, ensuring people are supported to have the knowledge, skills and expertise to manage their own health and care is a powerful way of alleviating some of the pressure on the system while ensuring people are at the centre of their care.

Technology has already made health information more accessible, but now it can potentially go further to help those with low health literacy. The latest development in AI is a classification of tools called large language models (for example Chat GPT). These tools have the potential to make health and care information more personalised, understandable and actionable This supports patients to have greater understanding of their health conditions, have greater control over their health and care and be more confident and informed to take the steps they choose to in order to take care of themselves.

The enthusiasm for large language models is unprecedented – they are one of the most rapidly adopted technologies to date. Once you try using these tools the glimpses of potential are incredible. Recent research demonstrated that it’s possible to use these technologies to re-write medical consent information to a reading age of 13 from a university-educated level starting point while still keeping all the essential information.

'There’s also potential to be able to use an individual’s medical records to identify the most relevant self-care information which can then be re-written to be more accessible using AI.'

This powerful technology has the potential to provide individuals with tailored health information based on public and trusted sources at a literacy level that matches their capabilities, empowering them in their own care. There’s also potential to be able to use an individual’s medical records to identify the most relevant self-care information which can then be re-written to be more accessible using AI. But it doesn’t need to stop there. Large language models create text but are part of a broader category called generative AI – which includes AI that can create new images, audio and video. So, it is potentially possible to go one step further and generate tailored content in multiple different formats, like animations or audio information, by linking several different AI tools together. All this could reduce the literacy and numeracy barriers to self-care but to do so requires collective action by NHS leaders working with patients, staff and innovators to set a vision and direction in using AI to improve patient empowerment.

This is a new technology and there’s no shortage of potential hurdles that need to be mitigated, including development and regulation. AI like large language models can – and do – make things up, so having the right processes and improvement mechanisms in place is essential. The difference between reconstituting public information and clinical guidance has to be clear as the commonly used large language models are an unregulated consumer technology, not a medical device. These tools also share information with their supplier which could create significant privacy concerns. So NHS leadership, regulators and innovators need to come together with staff and patients to co-develop these technologies further, protect privacy and ensure acceptable clinical safety with patient benefit.

The potential for significant improvements to health and care service delivery through the latest AI developments is just one side of the equation. We shouldn’t overlook the potential for AI technologies to ensure people are at the centre of health and care services, with greater control of their care and are more empowered to self-care. Yes, these tools may improve how clinical services are delivered but a greater shift could be brought about if they are designed and developed so people are more in control of their own health, wellbeing and self-care.