📣 Digital Health Download: September 2026

0:02Welcome to the Heart of Healthcare podcast. I'm Halle Teko. I'm Michael Esquivel. And I'm Steve Krause. And every Monday, we bring you the latest in healthcare innovation as we sit down with entrepreneurs and industry experts. So buckle up and join us as we figure out how to improve healthcare for all.

0:32Hello, listeners, and welcome back to the show. Today is our monthly digital health download where we cover the biggest headlines in digital health. I'm Hallie Tekko flying solo this month. And fair warning, there was a lot of news, so I'm covering more stories than usual today, which means we're moving a bit faster through each one. Here's what's on deck. Betting on clinical trials, Meta's landmark settlement over social media and kids, AI outperforming doctors and making healthcare more expensive, what could actually save healthcare$1 trillion, lawsuits involving Aura and Selena Gomez's mental health startup, and a surprising new finding about GLP-1s and employment.

1:18Let's get into it. Let's start with some shocking late-stage capitalism. Prediction Market Calci has moved beyond politics and sports betting and is now allowing you to bet on clinical trial outcomes and FDA regulatory reviews. The prediction market industry hit$24 billion in monthly global trading this year, up from $5 billion last year. And to keep that gravy train growing, they have expanded into new categories like healthcare. Joshua Peterson, a Boston University professor whose 12-year-old son is in a cancer trial, called it a death market. The possibility that someone out there might be wagering against my son, hoping that the treatment fails so they can make money, is enraging.

2:08Former FDA Commissioner Robert Califf warned that turning loose a betting market in an ongoing randomized clinical trial really is a breach of scientific conduct. And researchers have pointed out that trial staff from pharmacists to coordinators could quietly influence outcomes for financial gain. The Kaoshi CEO argues that their product offers a new level of transparency to biotech. The odds that a drug will succeed are among the most valuable numbers in the economy, they say, but also the least visible. While banks and pharma companies have these estimates behind closed doors, they believe CalSHI can make these predictions visible to all.

2:58A CalSHI spokesperson also pointed out that betting on clinical trials is already happening today on the stock market. I have to be honest, I've never used CalSHI before, so I took a look around the website at some of these bets that are currently open. There's one called Will the FDA Approve Reditutride? This is Eli Lilly's new triple agonist drug for obesity. 91 % say yes, this will happen before January 2028. There are also people betting on the number of cyclosporiasis cases this year in the US, which countries will report Ebola cases and if we'll have a pandemic this year. To me, this just feels super icky.

3:47We've spent years talking about financial incentives, distorting health care, and I'm not sure that adding literal bets on who gets sick and which drugs fail is the answer. Speaking of icky, last week, Meta agreed to a landmark settlement paying up to$18 billion and agreeing to make sweeping changes to Facebook and Instagram to address claims that it designed addictive features that harmed children. The deal resolves a multi-state consumer protection lawsuit brought by 47 U.S. states. The lawsuit alleged that Meta knowingly built addictive features into its app that encouraged compulsive use by minors and that exposed young users to serious mental health harms, including showing them content tied to eating disorders, self-harm, and suicide.

4:42For the first time ever, the platform will be required to make certain safeguards, including limiting daily time limit to two hours for kids, nighttime blocks, stronger age verification procedures, and much, much more. My friend and former business school classmate, Frances Haugen, was the whistleblower who leaked internal documents that originally sparked the investigations. We had her on the show in 2023 in an episode titled The Algorithms Fueling a Mental Health Crisis, where she talked about her work at Instagram and how social media algorithms impact public health, youth safety, and mental well-being.

5:22A movie about her journey as a whistleblower is being released in October by Sony Pictures called The Social Reckoning. I also highly recommend the book, Careless People, a memoir by Sarah Wynn-Williams, a former public policy director at Meta, about how leaders of the company prioritize global dominance and profits over human consequences. These two stories, Kalshi and Meta, are connected. Technology has externalities, and increasingly, those externalities are becoming health problems. We built social platforms designed to maximize engagement, and now we're dealing with the mental health consequences of compulsive use.

6:05We built frictionless ways to bet on almost anything from your phone, and now gambling markets are reaching into health care, not to mention fueling a worsening public health crisis with millions of Americans facing severe addiction and rising rates of depression and financial ruin. There's a quote I think about a lot from my husband, Jeff Hammerbacher. He worked at Facebook a long time ago. And in 2011, he told a Bloomberg reporter, the best minds of my generation are thinking about how to make people click ads. And that sucks. At the time, that critique was mostly about wasted talent. But I think the stakes have gotten even higher.

6:46We're starting to understand the real effects that these products can have on our health and well-being. And there's an interesting opportunity for digital health on the other side of this. I actually recently met with a founder, a company called Birch's Health, which provides virtual treatment for people struggling with gambling addiction. Technology can create entirely new vectors for addiction. And then another group of entrepreneurs has to build technology to help people recover for it. Maybe getting more of those great minds working on human health isn't such a bad outcome. Okay, moving on to the high and rising costs of healthcare.

7:24PwC surveyed 27 major health plans covering more than 103 million members and projects that commercial healthcare costs will jump 9 % in 2027, which will be the sharpest increase in nearly two decades. The reason? 70 % of health plans point to AI tools used by providers among their top three cost drivers, saying that AI-powered documentation and coding tools are capturing more billing complexity per visit. A UCSF study cited in the coverage found that AI adoption at practices correlated with more billing units per encounter and a modest bump in visit volume, with no increase in claim denials, meaning the extra billing is sticking.

8:11Other drivers they noted include pharmacy costs, more than 85 % of plans point to GLP-1s and specialty drugs, hospital consolidation, which gives providers more leverage in reimbursement negotiations, and behavior health utilization, which is up 62 % since 2018. A New England Journal of Medicine catalyst piece from several authors, including Bob Kocher at Venrock, similarly pointed out that AI is much more likely to raise total health spending than lower it, at least in the near term. AI scribes and tools free up physician time, but that time tends to translate into seeing more patients and ordering more tests and referrals and not lowering costs.

8:59They also point out that administrative savings from AI tends to get absorbed as profit margin rather than passed on to patients as lower prices. The exception is value-based or capitated payments, where a provider is paid a flat amount to keep their patients healthy. Their AI-free capacity can go towards higher value preventive care instead of more billing. The authors concluded, without significant changes in the healthcare system's financial incentives and market structures, AI will not slow cost growth. Said another way, AI can make doctors more productive, but in a fee-for-service system, that might just mean more healthcare spending, not less.

9:45The technology itself is neutral. It's the payment model that decides whether AI saves money or spends it. So how can we save money in healthcare? A new study out of the Yale School of Public Health, led by epidemiologist Allison Galvani, modeled what a single payer system like this would look like for healthcare spending and mortality. They project over$1 trillion in savings a year, which is roughly 20 % of current spending, and about 114 ,000 fewer deaths annually from having a single-payer system. The savings come from five places, lower drug prices, paying providers Medicare level rates, less administrative overhead, less fraudulent billing, and fewer preventable ER visits and hospitalizations.

10:41The model does add back roughly$304 billion for currently unmet medical needs and universal dental coverage. So this isn't just cost cutting, it's also expanding who and what is covered. As Galvani put it, healthcare costs have been rising faster than inflation and a staggering share of that spending is consumed by administrative middlemen. It's worth noting some limitations here. This is a model. It's not a live pilot, but we can look to other countries, most other wealthy countries in the world that have some form of a single payer system that both have lower costs than us in the US and longer lifespan.

11:25We keep looking to AI as the lever to bend the cost curve, but this is a reminder that the biggest lever might still be how we structure payments in the first place and not the technology layered on top of it. But there is one way that AI might be able to directly make healthcare deflationary, and that would be to drastically reduce labor costs like what we're seeing in the Doctronic pilot in Utah, where AI is helping with prescription renewals. So not just assisting clinicians, but I hate to say it, replacing the work that they're doing. Venture capitalist Vinod Khosla has been arguing that AI will replace doctors.

12:06He said this at one of the early Rock Health Summits maybe 15 years ago and got tons of pushback, but lately it seems that he might be right. Vinod recently partnered with oncologist and bioethicist Ezekiel Manuel and his son Neil Kostla, a healthcare AI founder, on a new JAMA viewpoint piece. By the way, both Vinod and Zeke and Bob have all been on this show. It is always good to go back and listen to past episodes. Anyhow, their new piece argues that autonomous AI, meaning AI operating without a physician, could soon outperform both physicians and physician AI teams on key cognitive tasks like gathering patient information, forming a diagnosis, and managing chronic disease.

12:53They cite Google AI's tool hitting 97 % accuracy on gathering patient information versus 50 % for physicians. And ChatGPT correctly ranking the right diagnosis first in 60 % of complex cases compared to under 16 % for a group of 20 internal medicine physicians on a similar test. Their most provocative claim is that adding a physician back into the loop actually makes outcomes worse once AI alone is already outperforming humans. And they point to over 100 human AI experiments and dozens of clinical studies to back that up. Of course, it is worth acknowledging that at least two of the authors have a direct financial stake in this argument.

13:40Neil Kostla runs an AI virtual care company and Vinod Kostla's firm, Kostla Ventures, invests heavily in AI healthcare startups. The American Medical Association pushed back hard. CEO John White said, medicine is not a collection of tasks. Getting the diagnosis right matters. Recommending the right treatment matters. He argues that the cited evidence comes largely from simulated tasks and not real clinical encounters. and that medicine requires judgment and epathy that those benchmarks don't capture. It's also worth noting that the AMA has some of its own bias too. It is a physician trade group that has spent decades and millions of dollars lobbying and fighting scope of practice expansions for nurse practitioners, PAs, pharmacists, all in the argument that only physicians are qualified to lead care.

14:36None of that makes White wrong, but the AMA has just as much of a stake in physicians staying central to care as Khosla has in AI taking on more of it. Moving on, Aura is getting sued over its sleep data. A proposed class action lawsuit accuses Aura of overstating how accurate its$300 plus ring tracks sleep stages. The suit claims that real sleep stage detection requires scalp electrodes and eye sensors, the kind only found in a clinical sleep lab and not on a finger worn ring and alleges or a stage estimates amount to a coin flips chance of being correct. Aura has advertised sleep staging accuracy as high as 95 % compared to clinical sleep labs after previously citing a 79 % figure.

15:31Companies get sued all the time and allegations make for good headlines. That last part is exactly what Tom Hale shared on LinkedIn. He calls the claims misinformation and says they stand by their science, research, and advocacy claims. So what's likely to happen? And cases like this rarely go to a full trial, even though the company says it's prepared to defend the case in court. The more probable path is a long pretrial fight over whether the 95 percent claim would mislead a reasonable customer, followed by a settlement that would likely include clear marketing language, possible in-app disclaimers and maybe some limited credits or refunds for some buyers.

16:14Or as best defense is that it's 95 % figure refers to sleep versus wake detection and not for stage classification and that the ring is not a medical device. The plaintiff's best leverage is independent research showing much lower stage accuracy and evidence that shoppers relied on the headline number. As an Aura customer, I think I'm on my third Aura ring. Let me tell you that I was not shopping and relying on this headline number. I think most users understand what they're getting. We can expect months of motions, a class certification battle, and then pressure on both sides to settle before Discovery exposes any sort of internal studies or ad approval records.

16:59This feels like a very annoying distraction ahead of their upcoming IPO. So speaking of lawsuits, two investor entities are suing Selena Gomez, her mother, Mandy Teefy, and the mental health platform they co-founded in 2021, Wondermine, seeking to recover more than$1 million. To be honest, I didn't even know they had a mental health startup. I went to the website and it seems more like a blog, podcast, newsletter, unclear how they plan to monetize this thing. but the complaint alleges that Gomez's abject dereliction of her duties as the company's head of marketing and claims that investors were kept in the dark.

17:40As the lawsuit puts it, for three years, while the company quietly collapsed around them, not one of its founders, officers, or directors said a word to the investors. Investors also allege promised partnerships with JP Morgan and Fidelity never materialized. The co-founder Daniela Pearson left the company in 2023 after clashing with Selena's mom, Tifi, according to reporting on the case. This isn't the first time questions about who was actually running Wondermine have surfaced. A year before this lawsuit, the cut published an investigation based on interviews with former employees who described Tifi running day-to-day operations largely on her own with limited financial oversight, while Gomez stayed pretty unengaged from the business itself.

18:28One former staffer put it, Selena knew her mother was not well enough to be running that company. Gomez, Teefee, and One or Mine have not publicly responded to allegations. Running a startup is really hard, way harder than raising money. I believe Selena likely had good intentions in starting this company, but she probably just didn't have the capacity or know-how on running a company. Last story for today, let's talk about GLP-1s. We've heard all about the benefits of these drugs from weight loss to heart health to curbing addiction. And now a new study is tying use of GLP-1s to employment benefits, specifically for women.

19:14Economist Rebecca Diamond, publishing through the National Bureau of Economic Research, found that unemployed women who started GLP-1 weight loss drugs saw their employment rate jump 27 % within just 18 months on the drug. That is a bigger swing than the employment gap between women with a high school diploma and women with a college degree. Diamond compared jobless women starting GLP-1 treatment to similar women who hadn't started yet. Single women on the drugs were also 29 % more likely to be married or cohabitating after 18 months. One important caveat here, women who already had jobs, unfortunately didn't see a bump in salary or promotions from starting the drugs.

20:02This effect is specific to getting hired in the first place. Diamond's read is blunt. This looks like an appearance based hiring bias and not really new skills that you're gaining while on the drug. She pointed to the equity problem directly, calling it a case of using medical treatment as a possible remedy for discrimination and noted that early adopters tend to be more financially advantaged to begin with. So is this good news that a drug can undo bias or unsettling news that hiring bias is strong enough to require a drug to undo it? Probably both. If GLP ones show how much bias still shapes who gets ahead under our current system, this story really asks a bigger version of the same question, who benefits, who gets left out under how we structure healthcare payments altogether.

20:56So much going on this month. Just one more tidbit I wanted to share that made me laugh. Bob Herman, a reporter at Stat News, reached out to the UnitedHealthcare team for comment on a piece he was working on. Later on, someone accidentally copied him on some internal deliberations where a UnitedHealth official said, we received an inquiry from our least favorite reporter from STAT, which honestly might be the highest compliment you can get as an investigative healthcare reporter. That is it for this month's Digital Health Download. Thank you for listening, and we'll see you next week.

21:41Thanks for listening to The Heart of Healthcare. If you enjoyed this episode and you'd like to support the podcast, please leave a rating and review. And don't forget to subscribe. The Heart of Healthcare is produced by Hallie Tecko and hosted by Michael Esquivel, Steve Krauss, and Hallie Tecko. The show is engineered, edited, and mixed by Kyle Moore. Visit our website, heartofhealthcarepodcast.com, for show notes and details. Yeah.