Can Artificial Intelligence Solve the Caregiving Crisis?

U.K.-based MiiCare has created a virtual digital health coach named Monica that’s inside a compact cube-shaped box typically placed in the living room. This MiiCube box gathers data from a gaggle of third-party sensors, connected devices and wearables to detect someone’s blood oxygen level, hydration and sleep. It also monitors when a person enters or leaves the home, goes to the bathroom, takes medications and so on.

Insights from the data can be shared with caregivers and medical providers. Monica herself greets the person when the person first walks into a room. She’ll remind them about appointments, exercise or meditating and is there to summon help if the person feels dizzy or falls. The company, which hopes to launch through partners in the U.S. in 2024, is running pilot programs in the Boston area and San Francisco.

Meanwhile, some of the AI tools hitting the market can help consumers make sense of the notes they receive after doctor’s visits. In theory, you won’t have to call a human to explain them.

“The most important thing that will be helpful to us moving forward with AI is the degree to which the tool we’re interacting with has access to information about our health,” Orlov says. But “if what you want to know and deploy in terms of AI depends on revealing what you personally consider to be your private information, you need to give permission, [and] you need to be asked.”

Indeed, concerns around privacy and the public’s lack of trust are among the reasons some people remain skittish about AI.

“As data scientists, sometimes there’s a tendency to push boundaries,” says Kelvin Summoogum, MiiCare founder and chief executive. “We all need to have this moral compass to know that anything we do has to be ethical.”

Founder and CEO Sam Yang of Xandar Kardian agrees that ensuring privacy is critical. But he also thinks it is important to highlight another key word: dignity. Cameras and wearables can rob older adults living alone of their dignity because these things “are a constant reminder that they’re weak and need help.”

AI shows promise to help caregivers figure out what’s next

If experts can use AI to help assess what medications older adults have to manage, determine that they’re taking them, analyze their biomarkers instantly and see whether their meds are working — “if we can make all these things sync up, that would be a beautiful dream,” says Marcie Merriman, leader of cultural insights and customer strategy at Ernst & Young Americas. Biomarkers are health characteristics that are measurable, such as A1C in your blood, blood pressure and pulse.

“It’s not going to happen fast,” she says. Still, promising AI solutions are springing up.

The free Together by Renee app for Apple iPhones from Los Angeles digital health start-up SixD lets you use a smartphone camera and AI to mine potentially helpful data. An Android version is expected soon.

Using the app, patients or caregivers can snap a picture of a pill bottle and have Together extract the drug name, doctor who prescribed it and pharmacist. At the appropriate time, the app can use AI and text-to-speech technology to call the pharmacy and order a refill.

Among other capabilities, someone can take a picture of the appointment card from a doctor’s office and have the app automatically add it to a calendar. The patient will receive a reminder for their next visit, whether it’s for a blood test or an electrocardiogram.

AI can analyze a selfie

What’s more, Together chief executive Nick Desai claims the app can measure a person’s heart rate, blood pressure and other vital signs just from a selfie and be as accurate as devices you might buy over the counter in a drugstore.

“We’re not going to be your doctor,” he says. “We’re just going to say, ‘Hey you’re taking a blood pressure medicine. Your blood pressure doesn’t seem to be going down. You should talk to your doctor about this.”

Consumers can also take a picture of health insurance cards to have the AI explain a policy’s benefits in easily understandable language, he says. A recently added feature promises to turn a 15- to 20-page document of medical instructions from a doctor into a quick summary of action items.

Wes Donohoe, chief executive and cofounder of a start-up called Helpful that integrates medical and insurance records and caregiving guides, says his company is a few weeks away from a generative AI model that will let someone ask open-ended questions along the lines of “My dad had a heart attack and bypass surgery. What do I do next?” Helpful will structure a personalized plan after having the AI pore through benefit options, doctor’s instructions and other information, he says.

AI can manage a caregiving social network

The idea behind many of these efforts is to personalize care. Just as everyone has a different set of apps on their phones, the AI services available to a caregiver would also differ, Moss of Care Daily says.

People can communicate with the AI caregivers at companies Care Daily partners with by phone, in text messages or through the TV, Moss says. But whenever possible he advises that AI caregiving be a family affair: “Family members who are being thrust into this caregiving role didn’t sign up for this, and they don’t know what to do. Part of the responsibility for an AI caregiver is to help educate the family about how to play this role.” Thus, the AI might arrange who needs to check in with Grandma on which day, whether that’s by phone, by sharing a picture or in person.

Moss also envisions AI caregiving as the “fabric for a new social network,” not something like Facebook, Instagram or TikTok, but rather one that brings families together privately for the sole purpose of providing care for someone.

Of course, not everyone has a trusted circle of relatives or community of friends who can help. That’s when a primary caregiver would need to summon outside care coordinators and clinicians.

AI avatars may be more empathetic than humans

In some instances, AI is helping chatbots and other tools become more empathetic, Orlov reports. By asking a series of questions, a bot may be able to gauge how a person being cared for is feeling.

Tone seems to matter.

Victor Wang, founder and chief executive of Care.coach in Millbrae, California, says generative AI improves the user experience for adults who engage with an empathetic chat-based virtual avatar that the company has introduced.

Responses from the AI avatar, called Fara, may differ from what you would hear from a medical provider, often for the better. Tell your doctor you have a rash, Wang says, and he may say something like “Uh, just put some hydrocortisone on it, three times a day. Let me know if it’s still a problem within a week.”

The AI, which is not a doctor, might give similar advice delivered in a friendlier fashion: “ ‘Oh, I’m so sorry to hear that. I would be really concerned as well if I had that kind of rash.’ You read it and you’d be like, ‘Wow, what a thoughtful doctor,’ ” Wang says.

Other potential advantages: AI responses come faster than a human at the other end can type. And the AI has a superior memory.

AI can’t be relied on alone

Wellthy, a New York City company that matches care coordinators with people who may be chronically ill, disabled or merely aging, is using generative AI to cut down on jargon and improve the tone of messages sent to members, says Kevin Roche, a cofounder and chief technology officer. But Roche is mindful of one of AI’s current flaws: hallucinations or answers that AIs spit out that sound plausible but are wrong or made up.

“We’re not having a chatbot where a member is just talking to an AI,” Roche says. “That’s too risky at this stage for high-stakes situations like caregiving. We want to make sure that everyone understands that you can’t take what [an AI] comes back with at face value. It always requires human review. … And it’ll be our stance that it should always be clear whether this is a message coming from a system or a human.”