Wearables and AI move deeper into care, but the evidence lags

On 1 October, Science published a special issue on women's health edited by an all-female team: Yevgeniya Nusinovich, Sacha Vignieri, Joana Osório, and Sarah Ross. According to STAT, they cited political attacks on women's rights and health care, along with funding cuts to research in the area, as urgent motives.

It is a useful frame for a week of digital health news that mixed solid data with thin pilots and a lot of vendor enthusiasm.

The issue's reviews argue that women experience functional pain at about twice the rate of men. They note that sex hormones, immune-neural interactions, the gut microbiome, and ovarian hormones all shape pain mechanisms, while research into female pain remains underfunded.

"Women are desperately waiting for concrete answers that might help eliminate the common 'all in your head' approach to their illnesses," the review's authors wrote, as quoted by STAT. A separate review argues that XX and XY chromosomes affect aging and disease so strongly that sex chromosomes should always be considered when developing precision medicine strategies.

Twenty people, moderate to large improvements

The same week brought a small but genuinely novel result on diet and mental health. A pilot study from UC San Francisco, published 23 September in JAMA Psychiatry and funded by the National Institutes of Health, followed 20 people with depression who cut their intake of ultra-processed foods by more than 85 per cent, according to ScienceDaily.

Participants showed moderate to large improvements across three different measures of depression, and anxiety scores improved modestly. The design was a crossover: one group stayed on its usual diet for four weeks while the other switched, then the groups swapped. Participants had an average age of 44, 70 per cent were female, 14 also had anxiety, and everyone had a high BMI plus at least one metabolic disorder.

"Reducing the amount of ultra-processed food in our diet can be challenging, because it's highly palatable, cheap, and convenient," said corresponding author Nyasha Chagwedera, an assistant professor in the UCSF Department of Psychiatry and Behavioral Sciences. "Ultra-processed food triggers dopamine, the pleasure hormone, which promotes craving."

Senior author Andrew Krystal said the pilot "successfully established the viability of carrying out this research in a larger-scale trial." Enrollment for phase 2 is expected to begin soon. Twenty people is not a treatment guideline, and the researchers say so themselves.

The money question nobody has answered

Meanwhile, the commercial side of digital health is moving faster than the evidence. Insurer startup Devoted Health announced nearly $1.2 billion in fresh funding after enrolling hundreds of thousands of new senior members, Endpoints News reported on 1 October. The same outlet reported exclusively that Parakeet Health raised $10 million to use AI for calling, texting, and emailing patients.

Those are funding rounds, not outcomes. A sharper question came from the payer side. The Blue Cross Blue Shield Association estimated that hospitals' use of AI-assisted medical coding contributed close to $1 billion, specifically $942 million, in additional costs for its health plans between 2023 and 2025, according to CNBC.

Roughly 70 per cent of that, or $653 million, was tied to additional diagnoses that were not accompanied by a change in care, Luke Chalker, BCBSA's senior vice president of product and data science, told CNBC. He stopped short of attributing the entire increase to AI.

Christopher Whaley, a health economist at Brown University who studies hospital coding, told CNBC that AI appears to be accelerating and making it easier to capture existing billing incentives. "In many cases, the diagnoses are legitimate and weren't captured," he said. Others, he added, "clinically just don't really matter and don't influence the patient's care" while still raising payment. BCBSA's report said the growth in complex coding came during a period when 60 per cent of hospital systems began using AI coding tools.

Data gaps, by design

If the US system struggles to measure what it pays for, it is also choosing to measure less about who it treats. In August, the CDC's National Center for Health Statistics proposed an overhaul of the National Health Interview Survey, which has collected Americans' health data since 1957, STAT reported on 1 October.

The redesign would cut the adult questionnaire from 482 questions to 150, removing questions about hearing aids, fatigue, cognition, and mobility supports such as wheelchairs and scooters. It would also send mailed questionnaires to roughly 25,000 households. Advocates say it would significantly undercount people with intellectual and developmental disabilities; a recent paper co-authored by CDC statisticians suggests past surveys may have missed 75 per cent of that population.

"If we don't have the government as a trusted source of data, asking questions about real people and what their real experiences are, we're never going to get better health outcomes," said Katy Neas, CEO of the advocacy group The Arc. A Department of Health and Human Services spokesperson declined to explain why the disability questions were struck, but indicated the changes were driven by a push to lower data collection costs.

Scott Landes, a sociology professor at Syracuse University, put it plainly: "You're making us choose between bad and worse."

Outside the US, diagnosis patterns are shifting too, which makes measurement harder still. A study led by the Barcelona Institute for Global Health and Aarhus University, published in JAMA Psychiatry and covered by ScienceDaily on 24 September, analysed more than 2.1 million children and adolescents in Denmark. It compared more than 71,000 diagnosed with ADHD or ASD between 2012 and 2022 against more than 713,000 without a diagnosis.

Children born with low birth weight were 54 per cent more likely to later receive an ADHD or ASD diagnosis at the start of the period, but only 17 per cent more likely by the end. "The key message is not that these risk factors are no longer important," said first author Magnus Elias Tarp. "What we found is that people diagnosed in recent years resemble the general population more closely than those who received the same diagnoses a decade ago."

That is a finding about who gets identified, not a verdict on what causes either condition, and the authors treat it that way.

Where the infrastructure money is going

The digital health build-out depends on compute and connectivity that few patients ever see. SoftBank completed its acquisition of digital infrastructure investor DigitalBridge for approximately $3.1 billion this week, Data Center Dynamics reported on 1 October. The deal takes the company private and de-lists it from the New York Stock Exchange.

DigitalBridge manages more than $108 billion in assets, with stakes in data centre operators including DataBank, Switch, Vantage Data Centers, and Yondr Group. Digital Realty, separately, is planning a cable landing station inside its LAX12 data center in El Segundo, California. Announced 22 September with Assured Communications, the project is aimed at Asia-Pacific to North America routes and expected to begin supporting cables in 2028, Data Center Knowledge reported on 30 September.

Both moves are bets that the demand behind AI in health and elsewhere keeps growing. Neither addresses whether the tools built on top of that capacity actually improve outcomes.

Elsewhere, the geopolitical picture is shifting as well. Chinese companies accounted for 44 per cent of the world's drone patent filings in 2024 against 17 per cent for American firms, though by value US drone patents were on a par with China's, according to a LexisNexis study covered by the South China Morning Post on 1 October. DJI held only 4 per cent of the industry's total patent assets, down from roughly 24 per cent at the end of 2018.

A separate Carnegie China study, reported by the same outlet on 25 September, found China employed 41 per cent of the world's leading AI researchers last year against 34 per cent in the US. This is a reversal from 2022, when the US led with 46 per cent against China's 27 per cent.

And on the ground in Germany, the digital paperwork keeps expanding: since 1 October, the official certificate of conduct can be issued as a sealed PDF delivered to the BundID citizen mailbox, at the same 13 euro fee as the paper version, though the paper original is still required for use abroad, heise reported. It is a small thing. It is also the layer on which everything else sits.

What the week actually shows is a pattern: the tools are arriving faster than the studies, and the measurement systems meant to catch the difference are being cut back.