Wearable AI and Digital Health Funding Surge as Depression Study and Claims Fight Collide

The study is small, and the researchers say so. Twenty participants, published Sept. 23 in JAMA Psychiatry and funded by the National Institutes of Health, moved from their usual diets to food low in ultra-processed products, then switched back after four weeks. According to ScienceDaily's write-up of the University of California, San Francisco release, participants cut ultra-processed intake by more than 85%. They showed moderate to large improvements across three depression measures, with a modest improvement in anxiety as well.

Andrew Krystal, the senior author and a professor in UCSF's departments of Psychiatry and Behavioral Sciences and Neurology, said the pilot "successfully established the viability of carrying out this research in a larger-scale trial." That is the honest reading: viability, not proof. Enrollment for phase 2 is expected to begin in 2028.

Why does this sit in a digital health wearable story at all? Because the same week has produced a run of evidence that the body's signals, whether from a gut lining, a wrist sensor or a billing code, are being measured more aggressively than the systems around them can interpret.

Twenty people, three measures, one hypothesis

Corresponding author Nyasha Chagwedera, who founded the UCSF Metabolic and Mental Health Program, offered a mechanism rather than a verdict. Ultra-processed foods may damage the protective lining of the gut, she said. This creates low-grade chronic inflammation that either causes depression directly or makes people more vulnerable to it. "In other words, junky processing may harm your gut in ways that ripple up to your brain," she said. She also noted the commercial problem: these foods are cheap, convenient and engineered to trigger dopamine, which makes them hard to quit.

The cohort was not a general population sample. Average age was 44, 70% were female, 14 of the 20 also had anxiety, and everyone had a high BMI plus at least one metabolic disorder such as hypertension, diabetes or elevated LDL cholesterol. That limits how far the findings travel. It also overlaps neatly with the population that consumer wearables and metabolic tracking apps are now chasing.

AI moves into the billing code

On the same day, CNBC reported a very different kind of health data problem. The Blue Cross Blue Shield Association estimated that hospitals' use of AI-assisted medical coding contributed close to $1 billion, or $942 million, in additional costs for its health plans between 2023 and 2025. BCBSA said much of the increase came from secondary diagnoses that moved patients into higher-paying reimbursement categories. About 70% of the total, or $653 million, was tied to additional diagnoses not accompanied by a change in care.

Luke Chalker, BCBSA's senior vice president of product and data science, told CNBC that "there is a clear disconnect between coding and treatment," but stopped short of blaming AI for the whole increase. "While multiple factors contribute to coding intensity, the findings suggest AI-enabled coding and documentation tools are playing a role," he said. He added that the costs can eventually show up as higher premiums and out-of-pocket payments. The finding landed alongside a forecast from benefits consulting firm Marsh that the cost per employee for health coverage is expected to rise 8.2% on average in 2027, the highest increase since 2003.

The American Hospital Association pushed back. A spokesperson told CNBC that patients are older and more clinically complex, and that AI tools help providers capture conditions for care planning. The AHA also said it was "particularly troubling to see insurers raising concerns about provider coding while continuing to rely on automated downcoding and denial practices." Chalker said Blue Cross Blue Shield companies use AI in claims review too, but that any clinical denial is always reviewed by a qualified human clinician.

Christopher Whaley, a health economist at Brown University who studies hospital coding, described the dynamic as an "administrative arms race." He said AI appears to be accelerating existing billing incentives. While many newly captured diagnoses are legitimate, others "clinically just don't really matter and don't influence the patient's care" while still raising payment.

Money keeps moving into the same space

Investors, meanwhile, are not waiting for the argument to resolve. Endpoints News reported on 1 October that Parakeet Health raised $10 million to automate patient outreach by AI phone calls, texts and emails. The same outlet reported that Medicare Advantage insurer Devoted Health raised nearly $1.2 billion after enrolling hundreds of thousands of new senior members.

TechCrunch's Equity podcast on 30 September turned to Neko Health, the body-scanning startup co-founded by Spotify's Daniel Ek, which raised $700 million. The episode noted that Midjourney is building its own body scanner and that Function Health has raised significant capital for a preventative-health platform. The question posed was blunt: is the scan worth it, and what would have to happen for body scanning to become part of the broader healthcare system?

That is the same question the UCSF pilot raises in a cheaper form. A wearable or a scan produces a measurement. Whether that measurement changes care depends on what the clinician, the insurer and the patient do next.

Digital sovereignty, and who can leave

There is a supply-side constraint too. Capgemini Research Institute surveyed 1,300 executives at large companies and public bodies. Il Sole 24 Ore reported on 1 October that 59% consider full digital sovereignty unrealistic, while 93% have discussed the topic at board level. In Italy, 94% have discussed it in the boardroom and 66% call full sovereignty unrealistic.

Monia Ferrari, Capgemini's chief executive in Italy, said the main weaknesses are a lack of visibility into dependencies, concentration of suppliers and the difficulty of replacing them. "Thinking you can eliminate all dependencies is unrealistic," she said, arguing instead for more autonomy where business risk is highest. The survey found 73% of Italian managers, close to the global 75%, put AI among their priority areas for digital sovereignty. The race to new models risks creating fresh concentration.

Only 49% of Italian executives said they would pay a premium for sovereign technology, and one in two would not accept slower performance or innovation cycles. That is the trade-off in plain numbers.

Context from outside the US and Europe

Rest of World published an interview on 24 September with Ruby Wang, a physician and consultant, about her forthcoming book on China's biotech and AI medicine sector. Wang said that in the UK and US, technology is "always an add-on in the health system," while in China the digital rails are enabled enough that society treats them as normal. She pointed to telemedicine, AI doctor avatars and drug licensing: China now accounts for about half of global drug licensing deals, and its share of global out-licensing deals could reach 60% this year.

Wang also drew limits. China is strongest in oncology and less mature in immunology, and its biotech companies are weak at commercialising, which is why out-licensing to partners such as Pfizer is common. She noted that Western rules including the Biosecure Act, the COINS Act and the BINSA bill could slow that flow. "Chinese digital health is not directly exportable," she said, adding that the West would never accept an AI doctor avatar.

Put the week together and a pattern appears. Sensors, scans, food diaries and coding algorithms are all producing more data about patients. The UCSF pilot is preliminary, the BCBSA estimate is disputed, and the funding rounds are private bets. None of that settles whether the extra measurement improves health, but it does show who is already paying to find out.