Can AI Bridge Gaps in Elderly Care and Support?

According to the 2023 Profile of Older Americans, a May 2024 report produced by the Administration on Community Living, a former department of the US Department of Health and Human Services, America’s population is getting older at a rapid pace.

In 2022, people older than 65 years represented a little over 17% of the population, a figure expected to grow to 22% by 2040. Among those people, close to 30% lived alone as of 2022. Given the health needs of this population — including those related to mental health — and the advancements artificial intelligence (AI) is making in the healthcare space, is it possible to apply AI to meeting some of the needs of older adults so that they’re better cared for and enjoy an increased quality of life in their later years? 

AI already is helping older adults, said Sumi Misra, MD, an associate professor of medicine at Vanderbilt University Medical Center and chief of palliative medicine at the Nashville Veterans Administration Medical Center, both in Nashville, Tennessee. They might not be aware of it, and those caring for them might not be, either.

“We already have been using AI in our patient population — sometimes we just don’t know it. The elderly are an incredibly rapidly growing population, this is a massive market,” Misra said. “But it (the development of AI) has to be done without exploitation.” 

While clinicians sometimes have mixed feelings on the efficacy of AI, one thing they can agree on is that older people should be able to have rich, purposeful lives. Shoshana Ungerleider, MD, an internal medicine physician and founder of the End Well Project, refers to it as the patient’s “health span,” as opposed to just their “lifespan.”

“When we talk about lifespan, we’re simply referring to how long we live — the number of years we’re alive. In contrast, health span asks a deeper, more human question: How well are we living in those years,” Ungerleider, who is also the host and producer of Before We Go and the TED Health Podcasts, said. “As a physician, I have cared for many people at the end of life. What I learned is that most of us aren’t trying to only achieve more time; we are hoping for time that feels meaningful. Health span focuses on quality: our ability to move, to connect, to think clearly, and to live in alignment with our values with dignity, agency, and presence.”

Physicians and the Use of AI in Senior Care: Practical Applications Have Already Arrived

Clinicians are already implementing the use of AI applications in the workplace in the same way that other businesspeople do: to speed up tasks and reduce workload. For example, AI note-takers dramatically ease the chore of charting, and the tedious work of coding is greatly simplified. Not only can this help providers avoid “pajama time” and thus burnout but also it can help cement the doctor-patient relationship.

“I know many clinicians who use these frequently, so that they’re not having to face a computer and type on a computer when they should be looking at the person who’s in their office,” said Jo Ellen Wilson, MD, PhD, an associate professor of psychiatry and behavioral sciences at Vanderbilt University Medical Center. “If it helps facilitate that type of connection, the more human-to-human connection, I think we should use these tools to help us regain that ability to help us connect with our patients across the room and not with the computer.”

In this scenario, the use of an AI tool could actually facilitate the very type of “health span” conversations to which Ungerleider referred.

“When we consider how we live — not just how long we get to live — we open up space to have good conversations: What does a good day look like? What brings you joy? What do you want your life to feel like, even when facing illness, grief, and advanced age,” Ungerleider said. “Health span allows us to move from a system of treating symptoms to one of treating the whole person — to live as well as we are able, for as long as possible. We think less about test scores and more about what is important to the individual.” 

Misra, who is the associate director of the Geriatric Research Education and Clinical Center within the VA Hospital in Nashville, Tennessee, said that AI is already being used to track these complex patients and their healthcare needs and to analyze their symptoms.

“When you think about the senior group composed of high-risk, high-care, high-needs patients, which is a lot of people who are frequent users of the healthcare system at multiple touchpoints, whether it’s the ED [emergency department], frequent calls or messaging to their primary care providers or geriatricians, people who get admitted often, we often have this Medicare data about people who spend 90% of their Medicare dollars in the last six months of their lives,” Misra said.

“So there has been a lot of health monitoring in this population anyway to see how we can improve their care, how we can improve their experience. There are things like Medicare Advantage Plans, nursing homes, chronic disease management programs, which already use health monitoring. They do sophisticated data analyses, they use a lot of predictive modeling, and they also use a lot of personalized interventions. Even the Apple Watch can do (basic) EKGs now.”

Ungerleider said that AI-driven medical tools including the ones Misra enumerated enhance physicians’ traditional risk assessment methods by exposing early warning signs potentially even before symptoms start.

“We may be able to intervene in ways that slow deterioration, improve outcomes, and allow more people to maintain their independence, longer,” she said. “This technology also allows us to have conversations based on values. When someone learns they are at higher risk for a condition like cognitive decline or heart failure, we have the opportunity not only for medical intervention but also to have a ‘planning’ conversation — to discuss who they trust to make decisions on their behalf, their notions of dignity, and how they want to live, not just how they want to be treated. Medical forecasting enables us to transition from ‘if’ to ‘when and how,’ enabling patients and clinicians to walk the path from crisis care to intentional and personalized care.”

Misra said that the capabilities and efficacy of these tools can be furthered by augmenting what’s available with the addition of more personal patient data. She gave the example of an AI tool that tracks the glucose level of a patient who lives alone being combined with sensors that monitor how many times a day the patient’s refrigerator and pantry doors open, and/or how many times an order is placed for online delivery, and aggregates all of that information. If the patient’s glucose level trends up and the patient persists in going about some preset combination of other behaviors, perhaps their emergency contact, or an emergency monitoring contact, is alerted to intervene before the patient is hospitalized for diabetic ketoacidosis.

However, the more personal information is collected, the more of a “watchdog” these applications become, the more this technology will necessitate further discussions about privacy and data security.

“Anything can have a beautiful and positive impact, but then there’s also the dark side of it,” Misra said.

The Role of AI in Caregiver Support and Assistance

Statistics from the same Administration on Community Living report mentioned above showed that family caregivers are the largest source of care and long-term support for older adults in the US, enabling their relatives to “age in place” as opposed to moving into a nursing home.

While a substantial body of evidence does suggest a correlation between the two, the key factor is not just staying in the house itself, it’s the wide range of health and social benefits that often come with it, including home-cooked meals and companionship.

More than 60% of caregivers say that their loved one would not be able to remain at home if it were not for the support they provided, so much so that a 2023 study from American Association of Retired Persons (AARP) valued the amount of unpaid labor provided by family caregivers at $600 billion annually.

The amount of time and energy these caregivers devote to ensuring that their loved ones can stay at home is causing caregiver stress to reach an all-time high. As reported in the 2025 edition of Caregiving in the US, a project of AARP and the National Alliance on Caregiving, 29% of caregivers are also raising families of their own.

A growing number of caregivers are providing high-intensity care, tasks for which most were never trained: 40% are doing things like administering injections or managing equipment. One in 5 caregivers report poor health; a quarter are taking on debt due to caregiving, 50% report negative financial impact due to caregiving, and 1 in 5 cannot afford basic needs like food. Seven in 10 family caregivers are employed, but many face disruptions and don’t have access to supportive benefits. The doctors said that AI can affect the lives of these caregivers with minimal impact to the care they are providing to the older adults in their homes.

“We can automate tasks for patients and their caregivers to decrease the burden of some of these caregiving tasks, beginning with organizational tasks, like managing doctor’s appointments, and legal documents,” said Wilson.

Misra said one thing that interests her is how AI applications can be personalized to enable older adults to keep their dignity and independence. AI applications could be useful in this situation even when the older adult is living with a caregiver, for example, in the instance of a person with mild cognitive issues who needs to be reminded to take their medications or perform other activities of daily living, but bristles when reminded by a son or daughter.

“Asking them ‘Have you showered?’ ‘Have you taken your medicine?’ — without them having to prompt (an AI device) to tell them back a particular way, that’s one thing that’s exciting,” Misra said. “How it can personalize care where they want to maintain their dignity and their independence, but it’s also a reminder in a thoughtful and not a punitive way. That, I think, is the really important part.”

AI’s Ability to Directly Assist Older Adults

Beginning with older adults who have some computer proficiency, Wilson said the technology can speed up tasks and make them more efficient, possibly also increasing an older adult’s confidence in some areas.

“Think about an older adult wanting to plan a brief vacation with her family, but they’ve never been to that location before, so she could use an AI software like ChatGPT or Gemini to help plan a brief itinerary, that could be quite helpful,” she said. “You can also automate certain tasks that are mundane or might take up a lot of time, so some sort of virtual assistant could also increase efficiency and automation. I think it can also improve decision-making in certain situations, so when you have limited information about, let’s say, a new diagnosis or something that you want to learn more about, it can increase the efficiency of that process.”

Wilson said that it’s important to cross-reference all results, as these systems are only as good as the information that they can access — which can vary dramatically. There are other applications that don’t depend as much on this kind of factual output that Wilson said can be very useful, though.

“Certain applications with AI tools can be helpful in a therapeutic sense from a mental health standpoint. There are really nice meditation mindfulness apps that use AI-based interfaces, and others that use cognitive behavioral techniques that can also be quite helpful, in particular for those with depression or anxiety, where maybe they have limited ability and cannot easily get to a clinician’s office on a routine basis,” she said. “I think we have to be careful as AI is evolving and its usefulness becomes more clear and more mainstream. We want to make sure that AI is not replacing authentic human connection. AI shouldn’t replace nuanced clinical assessment by an experienced clinician — you can’t replace the reasoning of an experienced clinician for medical decision-making.”

One interesting application in this area is the use of AI-based pets. Wilson said she has seen these used in the hospital.

“With critically ill folks in the ICU, these can stimulate what may be a familiar environment. Maybe they had a beloved pet years ago and that helps them be comforted or to connect with something,” she said. “Then, I think using it as a therapeutic tool can be useful.”

On the whole, however, Wilson remained skeptical about the efficacy of current applications in helping older adults.

Misra, who admitted to being a “glass half full kind of person,” referenced the use of “reminder” AI applications, prompting older adults to perform activities of daily living, as also appropriate for seniors living alone, and said that once they delivered this physical benefit, they could also foster a sense of competency among members of this audience.

“That plays into their psychosocial and emotional wellbeing, which is a big deal and a big issue for elderly people now. There’s way more depression, way more anxiety,” she said. “I’m kind of on the fence about the chatbots and the virtual companions — I think we still have a long way to go there.”

However, Misra doesn’t think there will necessarily be baked-in hesitancy from seniors with respect to learning new technology. In fact, Misra said she recently showed her 83-year-old mother how to use ChatGPT. Her mother took the opportunity to ask it for advice on how to deal with a couple of interpersonal issues and was impressed with the answers. Misra suggested that she personalize the answers by telling ChatGPT that she was older, so she typed in “I’m an 83-year-old woman,” and it quickly reformatted those answers to sound more like something someone her age would say.

“Now she’s become the guru in her retirement community for teaching people how to use ChatGPT,” Misra said.

Drawbacks and Red Flags

Anytime you’re using a search engine, you’re interfacing with AI to some extent. Wilson urged caution with AI, given its power and ubiquity.

“Some people don’t even realize that when they go to a common search engine like Google that AI is being used, oftentimes, to answer your questions,” Wilson said. “So there’s got to be a healthy degree of skepticism when you read the results. For example, what I’m seeing is that sometimes, patients come into my office and think they’ve diagnosed themselves with something or are concerned that they have a condition, when in reality, something else may be at play.”

Misra said that at 57 years, she feels like people at 75 or 80 years are seniors, but that even she can feel fairly incompetent in the presence of her 18-year-old son or her 25-year-old daughter and how agile they are with their phones. Older adults, like fish out of water, will be vulnerable in the field of AI.

“If I think of someone who’s older than I am, and hasn’t been in an academic center, not surrounded by a bunch of trainees, they are kind of isolated. They are susceptible to being scammed — they can be sold a bill of goods that’s almost too good to be true. This can be made to be sound fantastic and like a solution to so many things, but if you don’t have a basic understanding of how to use it, but if you have someone, or some agency, or some entity that has something to sell, you become this very large market that’s vulnerable to exploitation.”

Misra added that privacy and the safety of personal health information will both continue to be mounting concerns the more AI does for all of us in this area.

The Future of AI in Caring for Older Adults

Misra said that the biggest need is for the developer to catch up with the lives of older adults and address what she refers to as the digital divide. In the case of this audience, the divide is multi-faceted, requiring the consideration of many different dimensions.

“The question is how to bridge the digital divide for people who aren’t AI savvy, digital savvy, computer savvy, gadget savvy,” she said. “Then there’s the physical part of it, do you have the manual dexterity, or the hearing, or the speech — you know, think of someone with a stroke, who has some dysphagia or dysarthria, or someone like me, who may have a slight accent. Does that technology understand what you’re saying and the way you’re saying it? There may be medical issues that cause language issues, or there’s the international aspect.”

Wilson said that AI developers also need to ensure that they are not favoring younger, nonimpaired audiences when programming.

“Developers need to make elderly-friendly AI tools and prompts that are accessible by older adults, including those with major neurocognitive disorders like dementia,” Wilson said. “They also should make sure that they’re training their algorithms on older adults’ material, because with some of them, I feel like there are some biases and some of the algorithms where it’s not necessarily as applicable to older adults.”