...

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.

This is a preview of the September 17 edition of Access Health—Tap here to get this newsletter delivered straight to your inbox on Thursday mornings.

Good morning. I'm wrapping up my stage notes for Newsweek's AI Health Summit next week and noticing a difference from digital-health-care-forums past. Most of my questions are not about what AI can do, but about what people can do.

Read More on Access Health Newsletter

Who responds when an algorithm flags a deteriorating patient? What happens to a clinician's role when AI allows them to manage conditions that once required a specialist? If AI can draft a clinical trial protocol in minutes, can the contracting, budgeting and coverage processes around it move any faster?

AI can change what happens in one step of a workflow. Health care leaders now have to figure out how every step around it changes, too.

That was the theme I kept hearing in briefing calls ahead of the Summit, and it became the focus of a story I published this week. Health systems and technology developers have spent the past several years proving that AI can do remarkable things. But integrating those capabilities into decades-old workflows, job descriptions and decision-making structures is proving to be a very different challenge.

Currently, the fundamental questions in C-suites are decidedly low-tech: Who does what? In what order? Who makes the final decision? And what happens when the old way of working no longer fits? Low-tech, I'm told, does not mean less complicated.

The 2026 Stanford AI Index found that AI performance continued to rise sharply across advanced reasoning and agentic benchmarks in 2025. On SWE-bench Verified, which measures whether AI can independently solve real-world software engineering problems, performance increased from roughly 60 percent in 2024 to nearly 100 percent in 2025. On OSWorld, which tests agents performing computer tasks, performance went from roughly 6 percent to 75 percent over the same period. Frontier models also gained 30 percentage points in one year on Humanity's Last Exam, a benchmark deliberately designed to remain difficult for advanced AI.

In that same one-year period, many of us sat in boardrooms and conference halls, drawing on principles from undergraduate business courses and debating change management strategies that predate the HITECH Act. We'll still be talking about them in New York City next week, and we'll likely still be talking about them next year. Health care is a distinctly human business, and although humans built Rome, they couldn't do it in a day.

Dan Shoenthal, chief innovation officer at MD Anderson Cancer Center, believes health care leaders may need to get comfortable with that mismatch between AI's evolution and our own. Technology companies, he told me this week, are “running at an even faster pace than we are,” while the time it takes a health system to actually realize value from their products is longer than many vendors expect.

"You have to experiment to understand how a role needs to evolve," he said. "You can write it down on paper and plan all you want, but until you start to actually get your hands on the technology and experiment with it, you're never going to know.”

In fact, Shoenthal pushed back on the anxiety many executives share, worrying their organizations are already falling behind. Rather than racing to turn on the next tool because the health system across town has it, he wants providers to spend more time sharing what worked, what didn't and what “good” actually looks like.

“We're all in the same place,” he said.

I'll see you next week for a Pulse Check with Shoenthal and, likely, a few stubbornly human lessons from a day devoted to artificial intelligence.

(P.S., if you're a health care executive and would like join us for the AI Health Summit in New York City on September 23, email me at a.kayser@newsweek.com to inquire about a complimentary pass.)

In Other News

Major health care headlines from the week

• Newly released government records show widespread delays and technical problems in Medicare’s WISeR model, which uses AI-assisted reviews for prior authorization. The pilot operates in six states and covers 13 services CMS has identified as vulnerable to fraud, waste or inappropriate use.

  • The records, obtained through a Freedom of Information Act lawsuit by the Electronic Frontier Foundation, show some requests stretched far beyond CMS’ expected 72-hour turnaround. EFF said one request went unanswered for 83 days, while providers reported system outages, inappropriate denials and delays obtaining approval for medically necessary procedures.
  • The findings are raising questions about how thoroughly the AI systems were tested before deployment. CMS has said a qualified human clinician must review all denials and that it will take corrective action to ensure the model meets goals for timeliness, accuracy and transparency. WISeR is scheduled to run through 2031.
  • Get the full scoop at Newsweek.

• In other CMS news, the ACCESS model is expanding to cover heart failure, COPD, substance use disorders and nicotine dependence beginning in spring 2027.

  • ACCESS—the voluntary Original Medicare model that uses technology-supported care, including virtual care, remote monitoring, connected devices and wearables, between regular visits—now has 160 participating organizations, and major payers representing 165 million people across Medicare Advantage, Medicaid and commercial insurance have pledged to adopt an outcomes-based payment structure aligned with ACCESS.

• Memorial Sloan Kettering Cancer Center and OpenEvidence announced a partnership aimed at bringing precision oncology guidance directly into clinicians’ workflows. OpenEvidence will be integrated into MSK’s Epic workflow, while MSK’s OncoKB precision oncology knowledge base will be incorporated into OpenEvidence for clinicians outside the health system.

  • The integration will pair patients’ clinical context with MSK-curated interpretations of cancer-related genetic alterations. Within OpenEvidence and Epic, clinicians will be able to see whether a specific alteration is clinically actionable, its level of evidence and associated therapies alongside answers grounded in peer-reviewed literature.
  • The partnership also extends MSK’s precision oncology expertise beyond its own clinicians. OpenEvidence said more than half of U.S. hematologist-oncologists already use its platform; through the collaboration, all clinicians using OpenEvidence will gain access to OncoKB’s cancer-genomics interpretations.
  • Last January, I explored the promise of precision oncology, and the barriers keeping it from reaching more patients, in this cover story. Advanced genomic sequencing has made it possible to identify rare cancers and match some patients with highly targeted treatments, but until now, access to that expertise has largely been concentrated at major cancer centers like MSK.

• President Donald Trump’s third surgeon general nominee, Dr. Nicole Saphier, faced Senate questioning over her views on vaccines at yesterday’s hearing.

  • Saphier said she does not believe childhood vaccines cause autism and called the MMR vaccine “our greatest tool for combating measles,” the Associated Press reported. She also said the universal hepatitis B vaccine at birth is safe and effective.
  • Saphier would enter the role amid significant changes to federal vaccine policy. The Trump administration has moved to scale back parts of federal vaccine guidance, while medical groups have warned some changes could increase vaccine hesitancy. Saphier said she would seek to restore public trust if confirmed.
  • Saphier would still need approval from the Senate health committee and the full Senate. Her ethics filing shows investments in health care, insurance and other companies; she said she would divest holdings that pose conflicts and resign from businesses including DropRx if confirmed.

Pulse Check

Executive perspectives on key industry issues

...

Yesterday, I sat down with Bill Golden, former CEO of UnitedHealthcare's employer and individual businesses, for a rare treat of a conversation. Major payers are notoriously tight-lipped about their inner workings, including AI strategy and deployment. Now retired and several years removed from his executive role, Golden was willing to share what he learned from 25 years at the nation’s largest health insurer.

Golden joined Oxford Health Plans in 1997, stayed through its acquisition by UnitedHealthcare and eventually led United's roughly $60 billion commercial business. Today, his view stretches across the payer-provider divide. He advises health care companies, serves on provider boards and now works with Anterior, an AI company focused on health plan administration.

Anterior recently launched a campaign called “Beyond Faster Horses,” a play on the famous line attributed to Henry Ford about customers asking for faster horses rather than imagining the automobile. Ultimately, the campaign—fronted by health care leaders including Golden, former U.S. Secretary of Veterans Affairs David Shulkin, former Blue Cross Blue Shield of Minnesota CEO Craig Samitt and Stellarus CEO Vanessa Colella—argues that health care risks using AI to optimize processes that may need to be fundamentally redesigned.

Golden told me he sees that danger on both sides of the negotiating table. AI can make prior authorization, claims and payment integrity faster for payers, while simultaneously making revenue cycle more sophisticated for providers. But if the underlying relationship doesn't change, those advances could simply make each side better at fighting the other.

He also believes proper AI implementation could help break that cycle. By giving payers and providers better data on their respective pain points, he argues the technology could create a shared set of facts for more productive conversations about reimbursement and contracting, instead of the two sides continuing to, as he put it, “beat the crap out of each other.”

Read on for a portion of our interview.

Editor's Note: Responses have been lightly edited for length and clarity.

You spent 25 years at Oxford Health Plans and UnitedHealthcare, ultimately leading United’s commercial business. Now that you’re working with companies across the industry, how has your perspective on health plans changed?

When you take a step back, you realize, regardless of whether you are a fan of health insurance companies or not, they sit in a very powerful position within the health care ecosystem of the United States.

Every private equity firm I talk to, every board that I sit on, the first questions are: How would United react to this? What would United do in this situation? How do we approach a United or a Cigna about what we're trying to do?

It’s not like there are 50 of those companies. There are four, and then you have the nonprofit Blues [Blue Cross and Blue Shield health plans]. It’s not so much that they can dictate, but they certainly have big influence.

Most of my readers are hospital and health system executives, so they’re sitting on a very different side of this industrywide shift. What are you seeing in AI and health plans, and what pain points could the technology solve if deployed correctly?

There’s this huge complexity machine within the internal workings of a health plan. You have clinical, you have prior authorization, you have an intense amount of paper. You have the touch points of not only the clinical side and the provider side, but all of the surrounding ambulatory services.

There are tremendous opportunities within the administration and day-to-day workings of a health plan for AI to take a lot of the work that's done there and make it more succinct, accurate and efficient, so clinicians can spend way more of their time working on things that matter to members and customers.

It should make things more affordable and easier to understand. Whether it's an approval or a denial, claim processing should be easier and smoother. All of those things that affect members, clients and health systems are a huge opportunity for AI.

Where health care has really struggled over the years is that health insurance companies, really well-intended, said five or 10 years ago, “The internet's great. Apps are really good. Let's give members tons of information to become really excellent health care consumers.”

Sure enough, the information was really good. The detail was excellent. The problem is that people don't necessarily want to become great health care purchasers. They want people to make it easy for them. They want suggestions. Their point is: You already know. Tell me what you know.

When you interact with your health care company, whoever is paying your claims and working with you from an administrative standpoint, they should look at you not as one of thousands or millions of members. They should look at you as a person of one.

AI can take all of those things we just described internally and package them so that when I log on to my benefit portal, everything is about me and my family. It's not, “Here's a big, broad set of information and data. Go find the best doctor.”

That’s where AI can provide way more detail, way more information and make it such a better experience.

Health plans accounted for only a small share of health care AI spending last year. Why have insurers moved more slowly than health systems?

Name me the last big insurance company that got into commercial health insurance and said, “We're going to compete with United and Cigna and the Blues.” Almost none, at least from a scale standpoint.

There’s no incentive for health plans to say, “We have to play defense as much as offense.” That's one thing.

Number two, it's not a big-margin business. When you build your budgets in health care, you start with: Here's how much money I think we're going to make. Here's what we think trend is going to be. If trend is elevated higher than where we're pricing, you start looking at your budgets and saying, “What do I need to keep the lights on? How do I start skinnying down investments so that I continue to at least look like I'm making investments in health care technology, but I can't go all in because I have a lot of other things I have to take care of?”

I also think health care companies have looked at innovation as: Here's what we do today. Let's make it slightly better. Let's make it slightly better the next year. Make it slightly better the next year.

What you've almost never seen is a health care company say, “By the way, we're completely transforming our business, and we're going to wake up January 1 and be a completely different company.” It doesn't happen.

First off, it's a heavily regulated business, so you have to be darn careful around compliance and governance. If you start not paying claims appropriately or make a big mistake, the government gets involved. Your customers get involved. You have to be good stewards of the way you make investments and changes.

But incremental changes were always the safe place for health insurance companies. If you spent 10 years making incremental changes, you could look back and go, “We did great work for those 10 years. Look at how different we are as a business.”

I think the world is changing so quickly that 10 years becomes one year. If the big four and others, nonprofit Blues and others in the health care management space, don't make the AI investments now, they will be in a position where they're trying to play catch-up.

There's nothing worse when you're running a health plan with millions and millions of members than trying to play catch-up, because by the time you catch up, your competitor is already moving further away from you.

I think companies like United are putting big bets on AI to transform their businesses, and I think you're going to see some major changes over the next number of years.

UnitedHealthcare recently said it would eliminate nearly 30 percent of its prior authorization requirements. What does that suggest to you about where the business is heading?

I’ll take a step back on that one because I don't know that that's an AI play. I think that's: “We really took an internal look at how we conduct ourselves with our members.”

If you're approving virtually all of a certain category of prior authorizations, the question is: Why the hell do we do it? I think that's just logical. It doesn't make sense from a business standpoint internally to spend money on something you're ultimately paying for anyway.

I think that's more around: How do we serve the consumer? It's a reaction to the government giving them a hard time around how they conduct themselves in the market, the dissatisfaction you have with providers and the dissatisfaction you have with members who complain about denials.

However, they're still keeping a significant portion of prior authorization requirements. Why? Because they believe those actually add value.

I'm one of the defenders. I sit here and listen to people talk about denials, and I go, “Hold on a second. Insurance companies don't do denials because they like doing denials.” Most of the denials they do are on behalf of their customers who are actually paying the claims.

I can name zero clients that have told me, “I want you to pay for anything a doctor orders. Just pay. I'm just going to trust the doctor.” Not one client has ever said that to any health plan because they actually believe there is value in some of them.

A lot of prior authorization is around: Is it necessary? Is there an alternative way to get this service done? Is it in the most logical setting? Is it in a place where it's the best value for the ultimate payer of the claim?

I think artificial intelligence can help explain and provide data and information. If you're going to get knee surgery, it could say: “I just want to serve this to you. Look at this information. You can get your knee done here, potentially with the same doctor, or you can go to the most expensive hospital in the city.”

A lot of people, especially people who have deductibles and coinsurance and are participating in the payment, would prefer to go to the cheaper option. Artificial intelligence can help with that.

Tell me more about your work with Anterior. How would you describe its mission and the difference you think it can make in health plans’ AI deployment?

I think they are really ambitious. When I first took the meeting, I thought it was going to be, “We're looking to make the world slightly better and make it a little more efficient.” I thought, OK, that sounds like a cool company.

I realized very quickly that their ambitions are way bigger than that.

They want to improve the health care experience in this country. They want to start with health plans. They think there's the biggest opportunity to go from what they're doing to something that's way better, way more efficient and way more concise, and then utilize the things that they've built there to move within the health system and connect the bigger ecosystem.

It's not just taking something that's currently inefficient and making it more efficient. I think that would be a really boring business.

What Anterior is trying to do is take something that is inefficient, complex and not very consumer-friendly, fix that and then completely transform how that business can connect into all the other parts of the business.

If it was just, “We're going to take something that takes six weeks and make it three weeks. We're going to cut down your prior authorizations to a couple of hours,” those are all great things, but insurance companies have been working on that for years.

What’s exciting is taking the things they're looking at doing and then transforming the business. They're one of the few businesses out there that I think is willing to tackle that.

Is that ultimately about connecting health plans with other parts of the health care system, or connecting the different silos within a payer?

Ultimately, it'll be all of those things. But you'll have your hands full if you just spend your time within a health plan connecting all the different silos.

You have all the clinical business. You have operations, claims, payment integrity, collections and premiums. You have plan designs and marketing functions. You have actuarial, underwriting, legal and compliance.

There's a tremendous amount that they can do within that health plan ecosystem. Then, in conjunction with that, you start talking to health systems. You start talking to consultants and brokers around the types of things that you're doing and what you're trying to evolve to. You connect the business groups around how you want to share information and look at data.

There's tremendous opportunity for a company like Anterior within not only that internal ecosystem, but to really influence the external ecosystem also.

If payers and providers both get dramatically better at playing the existing reimbursement game with AI, have we actually improved health care? What has to change for AI to remove friction from the whole system rather than just one side?

You don't want the payment integrity bots to be fighting the revenue cycle management bots, and then all the bots get really smart and they just beat the hell out of each other, and nothing really happens.

What needs to happen, and where artificial intelligence can help, is that it could connect people and the data to say: “You guys are taking advantage here. We're taking advantage here. We need to get together and talk about how we contract.”

Why don't we all sit down and look at this information? Sit down with the big health system and say, “This is what the data shows. Why don't we put a contract around the idea that you get paid appropriately for the things that you do and should be paid for? And why don't we make sure we're not a hindrance to you getting paid, but we can look our customers in the eye and say you're not overpaying?”

Fraud, waste and abuse are real. I'm not accusing big health systems of fraud, waste and abuse, but even within a big health system, there are mistakes. That's just fact. And there are underpayments. That's just fact.

I now sit in the middle. I sit on boards of three providers, and I see where health care companies are super difficult in certain circumstances, and I struggle to defend them sometimes. I also defend them when I believe that they're actually doing things that are trying to save people money and they're well-intentioned.

The AI opportunity is to show: Here are the pain points on the payer side. Here are the pain points on the provider side. Could this lead to a more sophisticated conversation around a better way to contract, so that health systems know they have a partner that's going to pay them appropriately?

I think AI can help with that. You can construct appropriate contracts, performance-based contracts, doing the right things for members. If we can get to that level, I think there's a real opportunity for there to be a different dynamic than exists right now, which is: You beat the hell out of me, I beat the hell out of you. No one's happy. No one likes each other.

When I was running the New York market, I was really involved with the health care executives at NYU and Presbyterian. They're incredibly smart people. They're incredibly well-intentioned. They believe that their systems are the best. They're fighting tooth and nail to get them paid appropriately.

We beat the crap out of each other during debates on reimbursement. If AI were there to help with data and information, talk about denials and take all of the emotion out of this and really just get to the facts, I think there's a real opportunity for AI to influence those relationships.

We'll see, because you brought up a great point: Bots fighting bots is not a really good place for the health care industry to get to.

C-Suite Shuffles

Where health care leaders are coming and going

• Dr. David Skorton, president and CEO of the AAMC since July 2019, will retire from the role next June.

  • During his tenure, Skorton led the professional organization—and its 163 member medical schools—through the COVID-19 pandemic, developing a roadmap to right hospital staffing and equipment shortages and advocating respect towards the scientific process. Prior to leading the AAMC, he served as the 13th president of the Smithsonian Institution and president of both Cornell University and the University of Iowa.
  • The AAMC hired executive search firm Spencer Stuart to scout a successor.

• The health care AI company Notable tapped Dr. Nicholas Desai, a longtime Houston Methodist executive, as chief medical officer.

  • Desai has two decades of experience in health system C-suites, including a number of hospital leadership roles at Houston Methodist and his service as the enterprise’s chief medical information officer.
  • Notable is deployed at more than 12,000 care sites, according to a press release shared with Newsweek.

• UH Health—the health system of the University of Houston—named Beto López the inaugural executive director of its Institute for Digital Healthcare Transformation, launched in August.

  • López previously served on the founding executive team of UT Austin’s Dell Medical School, where he co-founded the Design Institute for Health and worked as its managing director. He also co-founded a social venture studio and venture capital fund focused on health care innovation.

This is a preview of the September 17 edition of Access Health—Tap here to get this newsletter delivered straight to your inbox on Thursday mornings.

View 0 Comments

Request Reprint & Licensing

Submit Correction

View Editorial & AI Guidelines