This weeks’ blog is by Dr Catharine Jenkins, Associate Professor, Dementia Care, Birmingham City University (catharine.jenkins@bcu.ac.uk). This blog is the second in our theme focusing on artificial intelligence.

Artificial Intelligence (AI), like Evidence-Based Practice, uses previous knowledge to expand on, reframe and develop new ideas. In this blog, catharine.jenkins@bcu.ac.uk, interviews ChatGPT, and interrogates it as to how  AI might impact on the lives of people living with dementia.

The term ‘AI’ refers to a computerised system which learns and communicates in a similar way to a person, via a screen or robot. There is limited evidence for AI in healthcare; generally we are anticipating innovations rather than exploring evidence retrospectively.

AI has however, already been incorporated in resources for people with dementia, for example in the famous ‘Paro’ Japanese seals which use animatronics and machine-learning to simulate presence of an affectionate companion animal, which learns its humans’ preferences to promote engagement through expression, sound and bodily movement¹. AI already analyses brain scans to promote quicker and more accurate diagnosis ². When asked how ChatGPT envisaged its input, it first suggested early detection, as ‘AI algorithms can analyse medical data, including brain scans and genetic information’. From nursing experience and evidence ³, I am aware that many early indications are observable and potentially measurable (using data such as videos, screenshots, photos), so could be included in AI assisted screening, saving money and stress, as only those at higher risk would proceed for scanning. A person’s gait, vocabulary, writing style, humour and facial expressions might be compared against previous behaviours. AI could analyse vast data sets, and through predictive modelling identify which factors correlate most significantly with cognitive changes.

ChatGPT noted concerns about ethical issues, although not around the impact of large-scale screening on individuals affected, their identity, the impact of stigma and potentially on health insurance and professional opportunities. As a machine, it can emulate human insights but not appreciate our more nuanced experiences. As AI is drawn from existing human-generated resources, it risks reproducing injustice like inequalities in access to resources (including the ‘digital divide’), inaccurate conclusions, bias in assumptions around cultural norms and further embedding existing healthcare priorities. This is illustrated in ChatGPT’s prioritisation of diagnosis and treatment decision-making over service-user and nursing concerns. ChatGPT recognised that humans are likely to become over-reliant on its skills, potentially so that ‘human expertise is overlooked, and care becomes impersonal and mechanised’. This warning alerts us to consider recommendations for nurses: It will be the human skills of sensitive, non-judgemental communication and ability to explain complex concepts clearly and simply that support people in deciding how and when they might add AI assistance into their lives. Nurses can lead personalisation of care and responsiveness to risk based on AI insights, advocating for equity in access.  Care could become more personal, less mechanised and for staff, more professionally satisfying.

People living with dementia need supportive interactions with others to maintain their identity and meet emotional needs for inclusion, occupation, comfort and ⁴. At first glance, these needs cannot be met through machines, which may ‘create perception of dehumanisation, where the personal touch and empathy provided by human caregivers are diminished, negatively impacting on dignity and psychological well-being’, as ChatGPT put it. However, people with dementia enjoy the company of robot seals, which create joy through their cuteness and responsiveness. Many older people are lonely and people with dementia often forget they have been visited, so feel increasingly isolated even with regular company. A robot which simulates presence could ease distress. It is also important to consider the impact of caring on family members, what is best for the person with dementia can be difficult and exhausting for their carer, who may neglect their own needs. The shortage of care workers means we cannot assume a human is available as an alternative to a robot – it is perhaps more likely that the person would be alone.

The most difficult interactions for people with dementia often occur when personal care is required; the sense of exposure and vulnerability could be alleviated using shower/bidet toilets, which gently wash and dry the person after use. Thus, technology has potential to take over aspects of healthcare that both carers and cared-for find problematic, enabling nursing time to be spent in humanising interventions that promote emotional well-being. Futuristic iterations of bidet toilets already examine urine to give information on health and could expand focus to analyse dietary needs. Similarly, existing technology supports care by prompting with medication according to feedback from worn devices, alerting carers to falls, monitoring vital signs and recognising patterns of risk. A person with dementia could use a tracking device combined with AI and robotics to facilitate independence and social inclusion. For example, an AI enhanced shopping trolley could lead a person from the local shop to a café and back home. If the person took an impromptu diversion, the tracking device could reorientate them or share their location. The role of nurses would change from advising family carers on strategies to minimise behavioural problems to having supportive conversations weighing ethical dilemmas around freedom and risk management when using AI enabled tools. Smart devices in the home could enable people to remain independent for longer. ChatGPT suggested ‘voice activated assistants, automated lighting and sensors that detect unusual behaviour’. In future, a personalised robot, integrated with a Smart home system, could provide a range of interventions that promote and monitor health, prompt a daily routine, provide comfort and fun, and enable safe social inclusion.

In navigating the range of options, people with dementia and their family carers will need informed advice, guidance and support from nurses, while nurse educators will need to support students in developing enhanced communication skills and AI-related understanding.

Technology is expensive, but, for example, as a care home costs around £1500 a week, a Paro seal and shower toilet could potentially pay for themselves by postponing care home admission. Future developments could be considered by evaluating financial and human costs and benefits, but co-created research should explore the preferences, experiences and concerns of people living with dementia in relation to AI and robotic-enabled support.

References

1. JENKINS, C. and GERMAINE, C. 2019. Living well in older age: What can we learn from the Japanese experience Nursing Older People 31 (1) pp.30-35

2. RANSON, J., BUCHOLC, M., LYALLl, D., NEWBY, D., WINCHESTER, L., OXTOBY, N., et al. 2023. Harnessing the potential of machine learning and artificial intelligence for dementia research. Brain Informatics. 10(1).

3. JENKINS, C., FELDMAN, G. 2018. Recognition of pre-clinical signs of dementia: A qualitative study exploring the experiences of family carers and professional care assistants. J Clin Nurs. 27 9-10_ 1931-1940 doi:10.1111/jocn.14333

4. KITWOOD, T., BROOKER D. Dementia Reconsidered Revisited: the person still comes first. Open University Press; 2019.