What are some ways AI could be incorporated into dementia care?
DeCamp: There’s a lot of potential to support caregivers for people with dementia or cognitive impairment. Some people have talked about carebots — AI robots that can help provide care for people with dementia. Those have been tried in a number of different countries, notably in Japan. But carebots obviously raise a whole host of interesting ethics questions about privacy, safety, choice, and so on.
Lum: From a health system perspective, AI chatbots could be used to support the information needs related to dementia care. Here at UCHealth, I have worked with the team that programs Livi, the chatbot for UCHealth, and we set Livi up to recognize and answer questions about memory and dementia. Livi can provide trusted, evidence-based dementia resources related to questions people have.
I am also the medical director of a new Medicare innovation program called the Guiding an Improved Dementia Experience (GUIDE) Program, and part of the national conversation is, what is the role of technology, or even specifically AI assistance, in care navigation for individuals with dementia and family caregivers? We are hearing that some GUIDE programs may be integrating the use of AI technology into the support of caregivers.
Perhaps an AI chatbot can provide that kind of support, but we need to study that. I've not seen strong examples of AI chatbots that can do that, so our team here doesn't use AI for our caregiving support. And we have humans who provide our longitudinal dementia care.
Could you speak to the connection between loneliness and dementia? Might AI tools such as chatbots help older adults be less lonely?
Lum: Loneliness is a big issue for older adults and family caregivers. There's increasing evidence that loneliness is a risk factor for developing cognitive impairment, so we want to find interventions that are scalable and that people can use if they are lonely or feeling social isolation. Older adults have a wide range of perspectives on their comfort with technology, broadly speaking, and so I imagine some individuals might be very interested in having an AI-facilitated experience, and for others, that's not of interest. I think we need to ask people what types of interventions and supports they would like and allow them to choose.
AI chatbots try to mirror human interaction, and I think they have great potential for older adults, though I'm concerned in the context of dementia. Dementia changes someone's perception of the world, and sometimes it changes language processing and verbal fluency. One person with cognitive impairment might really benefit from engaging with an AI-based technology, and might see it like an advanced “brain game,” and it could be engaging for them. On the other hand, if someone is having difficulties with insight, judgment, paranoia, or telling right from wrong, they may be less able to distinguish misinformation from a chatbot.
DeCamp: I've seen studies where loneliness improves after using a particular technology. And many people would say, well, that's a good thing. But it goes back to broader social questions. What type of society do we want to have? What problem are we trying to solve? If the problem with loneliness in our society has more to do with the way society is structured, and the burdens and demands prevent people from connecting with each other, these technologies may be a short-term fix to something that really demands a longer-term solution. It’s not always a bad thing to use technologies to alleviate loneliness. But we can't lose sight of the more fundamental causes of loneliness in society and continue to work on those.
Could AI help diagnose Alzheimer’s and other types of dementia?
DeCamp: There's a lot of excitement about that possibility. One example you often hear about is early diagnosis of Parkinson's disease. People are actively researching ways that AI can see subtle patterns in movement, speech, and so on that could be predictive of future Parkinson's disease, and likewise for dementia. For patients and families, the question is going to be, well, how good is it, and what can I do if I find out I have this condition? Or what if it is wrong?
Lum: AI comes in so many different shapes and sizes. Could an interactive AI-based chatbot ask good questions to get a sense of someone's functional ability, cognitive ability, and evaluate other medical conditions that might mimic dementia? If so, I could imagine it in the future. At that point, I think we're also talking, frankly, about AI replacing all clinicians. If it can diagnose dementia, that also perhaps means it can diagnose pneumonia and other things.
From the health system perspective, there are evidence-based algorithms that use natural language processing (NLP), which is a type of AI, to look at currently available electronic health records and help assess the risk of undiagnosed cognitive impairment, which is a helpful flag for me as a clinician to look at. Perhaps, through a patient portal message or the notes that are coming in from the nurse or the social worker, we may already have enough information to warrant a clinical evaluation. Sometimes, people don't feel comfortable bringing it up. It's not easy to talk about. So, there's certainly room for more supportive tools on the health care side.
What might be some of the risks of integrating AI into dementia care?
Lum: I would not want to see an AI tool be a replacement for supportive social relationships and community. If AI is augmenting brain health, that's great. But I don't want to see AI becoming someone’s primary relationship. I do think that that's possible in the context of dementia because someone can become quite fixated on a few things as their capacities get smaller. It’s common to have less ability to be motivated or to initiate something, and so you can be more centered on scrolling on a phone or perhaps interacting with an AI chatbot. And I think part of brain health is to really continue to initiate and challenge ourselves to have social connections in the community. There are lots of different benefits to social relationships.
And then I think there are the same potential risks that go along with any technology. Does everyone have equal access to something that works, that's technology-based? Are they tech literate? Do people have the availability of a care partner to help with use? We know many studies among individuals with dementia that telehealth can work, but often, the feasibility is supported through having an available supportive person, a family caregiver, and so individuals who don't have a family caregiver might not be able to benefit from the technology-based AI solutions. It's a chicken-and-egg situation. It could help someone be less lonely, but maybe they can't do it because they don't have a care partner who can set it up.
DeCamp: Consider AI-based home surveillance – technologies that take videos of the inside of a home continuously and that could detect a fall, for instance. That’s potentially really helpful, because we don't want people to have falls in the home and go unnoticed. Depending how it is done, though, people might perceive the technology as being mainly about surveillance versus supporting someone’s individual autonomy or freedom. How do we achieve that balance? Preserving autonomy is part of preserving an individual's dignity, especially when they have a chronic or serious illness. There can be a tension between the benefits that might come with technology versus respect for autonomy and the dignity of the person with the chronic condition. Work remains to get the balance right.
