Alexis Kayser

By Alexis Kayser

Healthcare Editor

0

Share

Newsweek is a Trust Project member

See more of our trusted coverage when you search.Prefer Newsweek on Googleto see more of our trusted coverage when you search.


Happy National Nurses Week! Now through May 12, several restaurants and food chains are offering freebies to show their appreciation. Click here for the rundown (and thanks for all you do, RNs!)


I'm writing this on the one-year anniversary of my role at Newsweek, and thus, the one-year anniversary of my deep dive into health care AI.

Read More on Health Care

To commemorate the occasion, I'll let you in on a secret: No topic has intrigued me more than AI, and no topic has made me internally groan more.

My most recent article on agentic AI is a prime example. While reporting this story, I spoke with physicians and executives at 10 companies developing AI agents. They often started the elevator pitch with some variation of this phrase: "With AI agents, you won't have to wait on hold to speak with a human call center representative. You can talk to the AI agent instead!"

Hmm. When I call the customer service line at Walgreens or Amazon or—heaven forbid—T-Mobile, the absolute best part of the experience is the part where I get to talk to a human. The rest of the time, I'm getting interrupted by the automated answering machine while I try to recite my birth date. I'm trying to punch my account number into the keypad before the robot tells me, "Sorry, I didn't get that." I'm yelling "I'd like to speak to a representative!" and growing increasingly frustrated by how not sorry the mechanical voice on the other end sounds.

This is the future that flashed before my eyes when I first heard that call centers are getting even techier. So I asked Ankit Jain, CEO of Infinitus—one such company that deploys AI agents to handle health care calls—to explain the difference.

"The conversational AI voice platform that we have built is extremely natural, extremely conversational," Jain said, "and it's akin to the ones that you would have if a human picked up right away."

Instead of saying, "Press 1 if you're a patient. Press 2 if you're a pharmacist," the AI agent might say, "Hello, can I know who's calling?" You can give an open-ended answer, and the system will generally know how to respond.

It isn't easy to achieve this level of fluency, Dr. Oliver Kharraz, CEO of Zocdoc, told me. Last week, the company launched its own AI agent, "Zo," to handle scheduling calls from patients (read all about it in this Newsweek exclusive).

The tool took two years to complete—in part because of linguistic challenges, Kharraz said. For example, Zo might ask a patient for their insurance card, and the patient may say to themselves, "Where did I put my insurance card?"

In an attempt to be helpful, the earlier iteration of Zo would suggest that the patient check their wallet or ask their spouse. But a human call center employee would typically tell the patient to take their time.

Zo was also trained in New York City, so when it first deployed in Texas, the agent was speaking too quickly and kept interrupting callers.

"We realized we have to adjust cadence, and we have to actually develop modules that [can determine] if the caller is likely done speaking or if they just made a pause," Kharraz said. Now, Zocdoc ensures that its models are trained on a broad geographic mix of speakers.

Across the board, AI developers (and health systems that use their agents) assured me that humans will still be accessible for more involved requests. It may be even easier to reach them without an AI agent—not an automated answering system—as our liaison.

What are your biggest questions or concerns about AI use in health care? Maybe I can find an answer. Email me at a.kayser@newsweek.com.


Health Care AI Playbook Graphic

...

Click here to register for free.


Essential Reading


Pulse Check

Miruna Sasu

...

Miruna Sasu is the president and CEO of COTA Healthcare. Its platform pulls data from electronic medical records, cleans and adjudicates the clinical notes, then returns them to health care organizations for research and development purposes.

You might recognize her name if you caught yesterday's story on AI use in the clinical trials business—we recently connected to discuss, as COTA uses an AI agent called CAILIN to sit atop data sets and answer questions about them via a search bar.

We also talked about the importance of real-world data—data derived from sources like EHRs, insurance claims and wearable devices—which can provide a more representative population. For this week's Pulse Check, I revisited my interview with Sasu to bring you some exclusive insights that didn't make it into the article. Enjoy!

Editor's note: Responses have been edited for length.

Clinical trials have not always been representative. When did the concept of real-world data start gaining traction in oncology, and how has that shift impacted research results?

We're not quite there on our journey to getting it where it needs to be. But I think it started with a revolution that [my former employer] Bristol Myers Squibb was a part of in 2016-2017, when folks were thinking about how to make clinical trials more diverse—how to match the patient population that's entering trials to the patient population in the United States and in the world.

Part of the reason to use real-world data is because we want to figure out where patients are, what their demographics are and how representative [those indicators are] of the population. You can't just enroll all men in a trial and expect that a woman, perhaps in her 60s, 70s or 80s, would take the same dose or react in the same way. Women have different hormones, different parts of the biological system that we need to account for, and so therefore, having representative [populations] actually benefits patients. We know if we need to tweak the dosing, or the way we administer the medicine, and whether it is safe for everybody.

Real-world data can show the journey of all the patients—not just some—that are selected for a trial. We have biomarkers for these patients, the data shows what is happening biologically. We have the ability to understand if they are traveling to participate in the trial or if they are native to the area. These are all things that influence a patient's ability to get to a treatment, stay on their treatment and follow treatment paradigm: So this is information that is invaluable to a life sciences company that's trying to make medicine that actually works and can be administered in the proper way and where the patient is.

Health system leaders that I speak with are really concerned about their data right now—how they're collecting it, how they're protecting it, how they're using it to guide care decisions and strategy internally. If they'd like to contribute that data to a larger cause, like oncology research, how would you recommend they go about that?

It starts with the technology. Most of the valuable information is in the doctor's notes. Finding a way to help document doctors notes outside of handwriting is an enormous step forward, because when we're putting technology on handwriting, we're going to be way less accurate than when we're putting technology on something that is already technology-first.

I think most docs would say, "Well, we kind of already do that"—but actually, there are lots of different little pieces of technology (that don't cost a lot of money) that can translate those doctors notes into data points that can help a company like COTA provide that information way, way faster and better.


C-Suite Shuffles


Executive Edge

Kristen Cerf

...

Kristen Cerf is the president and CEO of Blue Shield of California Promise Health Plan, a managed care organization offering Medicaid to more than 575,000 members across two of the most populous counties in California. This week, I asked her how she maintains her own health and wellness while managing a busy schedule.

Although our work responsibilities look very different, Cerf and I have a lot in common! We're both introverts who recharge by spending time alone or in nature (Cerf is fond of hiking, swimming, paddleboarding, skiing and snowshoeing). We also both have dogs who remind us to take at least two walks a day, and we find that our avid reading habits help us to relax.

"I encourage others to focus on what makes you feel calm, centered and brings your nervous system into alignment," Cerf said. Here's some more of her advice:

Before you go, take a deep dive into the "unprecedented fusion" of robotics and AI in this article by Dr. Majid Al Fayyad, CEO of King Faisal Specialist Hospital & Research Centre in Saudi Arabia and a member of Newsweek's CEO Circle.

This is a preview of Access Health—Tap here to get this newsletter delivered straight to your inbox.

View 0 Comments

Request Reprint & Licensing

Submit Correction

View Editorial & AI Guidelines