Withings USA expands monitoring into clinical care services.

Withings Medical Group will concentrate on cardiovascular, kidney, and metabolic health (CKM), beginning with hypertension, diabetes, and obesity. The medical group’s care team will work with the beneficiary’s primary care providers to build an individualized care plan for each member. The Withings Medical Group teams can also prescribe and adjust medications and support the beneficiary on lifestyle changes.

From the release: “This focused scope reflects a deliberate, clinically rigorous approach to entering care, and it marks only the beginning of a broader ambition for how Withings supports health over the long term. It also aligns with the ACCESS Model’s ten-year, phased approach to testing whether outcome-aligned, technology-supported care can improve outcomes for Medicare beneficiaries nationwide.”

In the UK, two Commons committees have urged the Health Innovation Minister to cancel the contract for the NHS Federated Data Platform (FDP).  The developer, Palantir, has come under fire in both the UK and US on its ties to defense and security agencies in those countries and with Israel. Palantir contracted with NHS England starting in 2023. The Commons committees objecting to Palantir to Health Innovation Minister Preet Kaur Gill are the Health and Social Care Committee and the Science, Innovation and Technology Committee. They are urging using a contractual break in February 2027 to term the contract, seek another contractor, or develop one in the UK. The concerned Members cross parties and are joined by the trade unions and 100 NHS data and technology specialists, who have separately objected to the FDP in a letter to Health Secretary James Murray.

The FDP is a key part of NHS digital transformation to facilitate storage and analysis of health datasets to improve planning, patient care and hospital operations. But the concerns apparently are more political, targeting Palantir and alleging that its defense and security contracts will discourage patients from allowing their medical information to be used. If we in the US shared similar concerns, Microsoft, Meta, AWS, and Oracle wouldn’t have a single healthcare contract. Computing UK

IKS Health wrapped up its TruBridge buy last Thursday for $557 million. The acquisition, announced last April, closed a quarter earlier than expected. IKS is a healthcare software company for administration and documentation, with TruBridge an EHR and revenue cycle management (RCM) provider to primarily rural hospitals. The combined company now has a US customer base of 150,000 clinicians and 2,000 healthcare organizations. IKS Health is traded in India with US operations in Dallas. The TruBridge acquisition is their third this year, financed with $600 million over five years in debt sourced from Citibank, JPMorgan Chase and Deutsche Bank. TruBridge shareholders received $26.25 in cash per share as offered in April. The release is singularly unrevealing on whether TruBridge will continue as a separate product line, nor on any operational/management continuity. It is another example of consolidation and ‘line extension’ between similar companies. MedCityNews

If it’s midyear, it’s time for a preliminary CMS Physician Fee Schedule draft. ATA ACTION has an early review of its telehealth and technology changes proposed by the Centers for Medicare & Medicaid Services (CMS) that affect physicians serving Medicare beneficiaries. From the release (PDF):

• Telehealth: In addition to changes to extend existing telehealth flexibilities, CMS proposes adding five codes to the Medicare Telehealth List, clarifying critical care consultation code descriptors, creating two new telehealth modifiers, and allowing physicians to bill for telehealth services involving residents when either the teaching physician or resident is in the room with the beneficiary.
• Remote Patient Monitoring: CMS proposes tightening guardrails and recalculating payment for remote patient monitoring (RPM) and remote treatment monitoring (RTM) services. Key changes include restricting RTM billing to patients with an established relationship with the billing practitioner, requiring a separately reportable initiating visit tied to the start of monitoring services, prohibiting use of third-party remote monitoring companies, lowering monitoring valuations, and potentially consolidating current CPT codes into four new G-codes (two RPM, two RTM) covering initial setup and monthly monitoring/management.
• Technology-Enabled Care: CMS is seeking broad feedback on modernizing Medicare reimbursement for technology-enabled care, with particular interest in primary care, care management, capitated payment models, and AI.
• New–Software as a Medical Service (SaMS): Aligning with the OPPS proposed rule, CMS introduces a new category – Software as a Medical Service – defined as algorithm-based clinical decision-support software with clinical or diagnostic functionality (excluding remote monitoring and prescription digital therapeutics). SaMS providing secondary analysis of lab data would shift from the Clinical Laboratory Fee Schedule to the Physician Fee Schedule.
• Merit-based Incentive Payment System (MIPS) Improvement Activities: CMS proposes two new technology-focused improvement activities: clinician use of AI to improve patient care, and use of interoperable clinical decision support.

Whether these survive to the Final PFS is dependent on comments, Congress, and budgets.

And now we have AI for our electronic toothbrush. Philips goes AI in its latest model, the Philips Sonicare DiamondClean 9900 Prestige. It uses built-in, on-device AI and spatially aware guidance to track brushing. There’s a 12-segment “Mouth Map” on the handle, with a light ring that gives real-time visual feedback on your coverage, pressure, and scrubbing habits. For those of us who like to press down, the SenseIQ sensor automatically adjusts intensity so we aren’t so bad. It has eight modes and three intensities, a bit more than their top-of-the-line existing Sonicares. And it comes in five premium colors! Well, well…as a faithful user of my Philips Sonicare toothbrush since a rather extensive dental  ‘remodeling’ a few years ago, and guilty of sometimes being a bit sketchy about my completeness, I am somewhat dismayed that to be au courant, I will need to acquire this new model…but it won’t be available till this fall in the US and Europe. Release, Mobihealthnews

Oracle is beginning to turn into one of those cautionary case studies. And that’s not good news for healthcare.

A recent analysis published in Seeking Alpha (may require free account, PDF here) will give one further pause.

The debt load is a crusher. There’s anywhere from $72 to $100 billion in debt on Oracle’s books, and that preceded the PIMCO financing of $16.3 billion that included equity. Their credit rating from major raters is one step above junk: BBB-negative from S&P Global Ratings and Moody’s Baa2 Negative outlook. [TTA 7 May] Oracle has already projected it needs more: an additional $40 billion in fiscal 2027 through a mix of debt and equity.

Net debt is soaring. Oracle ended FY 2026 with net debt of $97.6 billion. This represents a ~20% increase versus the $81.4 billion of FY 2025. Looking forward, this analyst is projecting that the net debt by end of FY 2027 will be in the range of ~$120 billion.

There’s another little problem called cash flow. Years of positive free cash flow, even during the pandemic recession, turned sour in their FY 26 to a negative $23.7 billion. The analysis states that Oracle is now heavily dependent on external cash–debt–to finance its AI infrastructure building. S&P also noted that capital expenditure has been geometrically increasing: $6.9 billion (2024), $21.1 billion (2025) and $55.7 billion (2026). The guidance for fiscal 2027? Between $90 and $95 billion. That is over a 70% increase. Cash flow clearly can’t pay down the debt they already have, a fact that banks factored in to cut off the lending spigot.

Its performance obligations are also crushing. $553 billion has been reported for OpenAI alone [TTA 7 May].  This article reports remaining performance obligations of $638 billion, about half of which is for OpenAI. Despite the disparity in numbers (perhaps completed obligations?), in this analyst’s view, something happens to OpenAI–for instance, a pullback in the confidence of markets to finance their aggressive expansion (and, this Editor would add, an IPO)–Oracle would be hit hard. To paraphrase an overworked analogy, if OpenAI stumbles and sprains its ankle, Oracle would break both legs.

But..but…isn’t demand supposed to be increasing, infinitely?  Increased demand is critical to continued financing. Yet there are leading indicators that demand is leveling off. Corporations are increasingly limiting internal use of AI by employees to tamp down soaring token usage costs and control budgets. These companies have also found that AI tools often do not work as well in replacing humans as they were cracked up to be. And now seemingly every software offering is ‘powered by AI’. The newness is wearing off.

The chip cycle is speeding up–not good news for AI infrastructure builders.  Nvidia is now in full production of its Vera Rubin chip, which is dramatically faster than the previous generation Blackwell chip. Our March reporting picked up that Oracle had forward commitments with Nvidia to use the older Blackwell chip for the OpenAI data centers, and OpenAI wasn’t happy about that. It’s not a simple matter of going to Nvidia and returning old chips and getting new ones–that isn’t done. Vera Rubin is also twice as expensive on a per-rack basis than Blackwell–$7.8 million versus $4 million. Add to this the long taffy pull of building (online is 12-18 months minimum), power, water, and labor costs…it adds up, as any builder will tell you.

Given the above factors, it shouldn’t be any surprise that Oracle stock is down 30% since last December and continues to decline.

Extra: The article also contains a short history of financial cycles. There are parallels drawn comparing the AI boom we are in to similar technology-driven boom and bust cycles, dating back to the building of canals in the 1700s, railways in the 1800s, and electrification in the early 1900s.  

Editor’s note: Any analysis of Oracle’s financial health is based on this article, previous articles, multiple sources, and Oracle’s disclosures. Our interest is primarily in the well-being of Oracle Health and the markets it serves. We do not engage in stock picking and none of the above should be construed as financial advice.

Bankrupt 23andMe finally settles with 2023 data breach victims

The benefits to those affected and filing a claim are:

• Up to $10,000 for Extraordinary Claims;
• Up to $165 for Health Information Claims;
• An estimated $100 for Statutory Cash Claims; and
• 5 years of Privacy & Medical Shield + Genetic Monitoring

This is a second round settlement. The first settlement dates back to before the bankruptcy [TTA 17 Sept 2024] and was for $30 million. 23andMe filed for bankruptcy on 24 March 2025 after multiple buyout offers by Anne Wojcicki and others. Wojcicki eventually bought the company through her TTAM Research Institute in July 2025, then converted the holding company name to Chrome Holding Co. and ChromeCo, Inc. d/b/a 23andMe.

What exactly Wojcicki intends to do with what’s left of 23andMe after emerging from bankruptcy is anyone’s guess.

The estimated number of people affected by the breach was 6.9 million. $14.29 million has already been distributed, possibly through the earlier settlement agreement. At the time, 23andMe said that $25 million was covered by their cyber insurance. Kroll authorized website, BBC News, Fox 5 Washington

The Neuralink brain computer interface (BCI) implant hits #26 in clinical trial, has first Canadian ALS patient. Lee Marten is a sergeant with the Vancouver police department who has had ALS since 2022. His mobility is limited to his hands operating his electric wheelchair. With the Neuralink N1 BCI implant, he can move a computer cursor by thought. Sgt. Marten is part of a clinical trial at University Health Network’s (UHN) Toronto Western Hospital. He is the 26th patient in a series of US and Canadian clinical trials among spinal injury quadriplegic and the first Canadian with ALS.  Mobihealthnews, CBC News

Resmed sells its MatrixCare software business for $490 million in cash. The buyer is private equity firm Frazier Healthcare Partners. Going forward, ResMed will use the cash to expand and concentrate on its core sleep health, respiratory, and connected home-based healthcare. The press release anticipates the closing as Q1 2027. (Resmed’s FY closes in September so this is end of 2026.) There is no information in the release on management or headquarters transitions. The Reuters report ​said that “transition ⁠services agreements with Frazier are expected to help ensure continuity across systems and operations and ​largely offset stranded costs in the first year ​after closing”.

The haircut was substantial. Resmed purchased MatrixCare in 2019 for $750 million for its EHR and software products to expand into a specialized area of post-acute care–an extremely tough one. They currently serve 15,000 providers across ​skilled nursing, senior living, long-term care, home health and hospice ​markets under the MatrixCare, Healthcare First, and Citus brands. However, its growth lagged company averages, according to KeyBanc Capital Markets’ note to investors. The sale at a clear loss has all signs of ‘throwing in the towel’ on a difficult to market and lower margin sector.

ResMed is retaining Brightree in the US and MEDIFOX DAN in Germany. Brightree provides business management and clinical solutions for post-acute care providers. MEDIFOX DAN provides management software for the German healthcare and social care sectors. Their individual value and profitability is unknown.

MatrixCare in the preliminary filing generated about $220 million in revenue and about $55 million in adjusted operating ‌profit ⁠in fiscal year 2026. Healthcare Dive

And to wind up on a lighter marketing note, Teladoc is now the preferred telehealth provider of the National Basketball Players Association (NBPA). The NBPA is the union for the professional players in the National Basketball Association (NBA). Teladoc will provide virtual care services for active and retired NBA players and their families. They will have access to Teladoc’s multidisciplinary virtual services, including primary care, 24/7 urgent care and preventive health screenings. NBPA and Teladoc Health will also host a preventive health screening for NBPA members at Summer League in Las Vegas. NBPA release, Mobihealthnews

One big years-long chute for Done Global’s Ruthia He and David Brody.

To be announced at a later date: when sentences will start and where they will be served, based on recommendations from the Bureau of Prisons.

While Done Global is effectively ‘done’, the company also does business under the name Mindful Health.

The points of the (at least) $100 million fraud were based on the illegal telehealth prescription of Adderall and other Schedule II stimulants such as Vyvance. Medications on Schedule II have accepted medical uses but carry high potential for abuse and psychological or physical dependence and thus are controlled:

• A scheme that used the Done Global technology platform, compensation structure, and clinical protocols to unlawfully distribute over 37 million pills of Adderall, defraud insurers of over $12 million, and obstruct the federal investigation that followed.
• The defendant (He) spent over $40 million on social media advertisements to deceive Americans into believing they had attention deficit hyperactivity disorder (ADHD), falsely diagnosing patients with ADHD, and distributing Adderall, including to patients who the company was warned were suffering from Adderall psychosis, bipolar, depression, anxiety, and other mental health conditions that were worsened by stimulant prescriptions. 
• These were designed to boost the Done Global valuation to above $1 billion.

From the detailed Department of Justice press release: “The defendants refused to hire or fired Done clinicians who did not participate in the conspiracy, while paying up to $60,000 per month to clinicians who signed Adderall prescriptions every 30 seconds. The defendants also used an “auto-refill” platform technology feature after an initial diagnosis to minimize follow-up appointments, where prescribers signed prescriptions for Adderall based on an automatically generated message that a patient desired a refill. Because of these policies, some patients went years without seeing clinicians, who continually authorized refills even through involuntary psychiatric holds or after the patients had died.”

Dr. Brody alone personally wrote prescriptions for 394,324 Schedule II stimulant pills prescribed to 6,559 Done members. He never evaluated them nor reviewed a single patient record. Part of the case was Done’s record of misdiagnosis, over-prescribing, and patient death.

Additional charges against Ms. He included diversion of company assets and operations abroad. As indictments neared in 2023/2024, they both deleted records, instructed employees to delete  incriminatory documents and messages from the company servers, and transferred communications to platforms such as WhatsApp and Signal using disappearing message settings to conceal sensitive information.

After completion of her sentence, it is likely that Ms. He will be deported to her home country of China. She attempted to flee to Hong Kong in February 2023 and was forced to surrender her passport. Despite this, she made a second attempt after obtaining Chinese travel documents, then was arrested and detained before trial as a flight risk. Two other factors were that she transferred $4.6 million in ad-related revenue to China and set up a shell company there. 

The Drug Enforcement Administration (DEA), HHS-Office of Inspector General, IRS Criminal Investigations, and the Centers for Medicare and Medicare Services (CMS) investigated as violations included financial diversion, record falsification, drug prescribing and Medicare/Medicaid fraud. The main DOJ units involved were the National Fraud Enforcement Division and the Health Care Fraud Unit.

The case was heard in the Federal Northern District of California by Judge Charles Breyer. Dr. Brody plans to appeal and significantly apologized for his actions. KQED and Behavioral Health Business.

Background on the indictment and conviction, TTA 24 January and prior as noted in the article. There is no additional information to date on the grand jury charges from December 2025 of the Done and Mindful companies. 

Editor’s POV: Done wasn’t the first–the far larger Cerebral was in 2022. While it is still in business, Cerebral has spent much of its time and fisc in litigation and settlements. Neither will be the last.  DOJ and Federal agencies are cracking down hard on waste, fraud, and abuse in healthcare; major targets of DOJ/HHS/DEA scrutiny are telementalhealth and substance use disorder (SUD) management, including prescribing and payments. More to come.

Another chute for Oracle, leading to the Hacking trap door. The vulnerability is within Oracle’s E-Business Suite (EBS) and affects the file transmission component of Oracle Payments payments. The flaw has been tracked as CVE-2026-46817 and carries a severity rating of 9.8 out of 10. 900 systems may be exposed, though Oracle flagged it in their May patch updates. The US National Vulnerability Database states that the vulnerability can be exploited remotely over HTTP without authentication. Oracle software seems to be a favorite target of hackers. Cybersec organizations Defused and Shadowserver, along with the US Cybersecurity and Infrastructure Security Agency (CISA) have flagged multiple software vulnerabilities across Oracle’s EBS, WebLogic Server, and PeopleSoft.  Computing UK

Is it a Chute or Ladder? Or Run For Your Life? OpenAI and Sam Altman made headlines before the July 4 celebrations with an offer of a 5% share of the company to the US Federal Government. Both OpenAI and Anthropic are imminent IPOs. The Financial Times report is based upon “early conversations” cited from two insiders. A 5% share, based on current valuations, is about $42.6 billion. It is not only a nice chunk of change in the public fisc but also a clever PR move that may help neutralize public blowback and downright hostility towards unwanted technology; sprawling, noisy, spewing, heat pooling and energy-greedy data centers; AI job displacement; companies discovering that AI is draining them dry without ROI; environmental and community groups; unions, local governments, and more. How it will mollify people who are angry about any of the previous is doubtful. Nor will it please those aligned with socialist Senator Bernie Sanders, who is demanding close to half of OpenAI’s and Anthropic’s value to be held in a sovereign wealth fund.

Whether Anthropic and the hyperscalers building like mad such as Oracle, Meta, Microsoft and others would follow Altman’s lead is debatable. The Computing UK take on this is that it is a cynical and obvious bribe, perhaps one worthy of a Marie Antoinette (who may never have said ‘let them eat cake’…but nevermind). The accountant or computing side of this Editor’s brain flags that neither OpenAI nor Anthropic are remotely profitable. Oracle as a hyperscaler has already fallen into a debt canyon from where it may not emerge. Likely, Oracle is not the only one either, if you isolate AI from hyperscalers’ other sustaining businesses.

A generous offer or a Trojan Horse? You pick….

In this Editor’s view, it satisfies no one, solves no real problems, puts power in exactly the wrong hands, and creates a major conflict of interest in the objective and responsible development of AI.

In the Ladder department, Meta is talking up selling its excess AI cloud computing capacity, thus creating a new revenue stream. If this Editor is not mistaken, it’s similar to the Amazon Web Services model. A second stream would be renting out its AI application programming interface (API) to developers. The charge will be based on usage. That assumes that Meta is envisioning a time that they will have that excess capacity to sell. Right now, there is a dearth of actual, online data centers and a shortage of capacity [TTA 14 May].  Computing UK

A $110 million split raise for Pearl Health rounds it out. The Medicare value-based care management services organization (MSO) and population health services for providers gained a $50 million Series C equity investment from Andreessen Horowitz with participation from Viking Global Investors, AlleyCorp, and Ulysses Capital, plus a $60 million debt facility led by Trinity Capital. The new funds will be used for developing their AI platform, turning clinical intelligence into measurable outcomes, growing health system and payer partnerships, expansion into Medicare Advantage, and new risk offerings. Pearl’s funding to date is $205 million since 2020. It claims that it reached profitability last year and will triple its patient base from 2024 to the end of this year. It currently serves 10,000 providers across 40 states, with more than 250,000 Medicare beneficiaries, in CMS ACO models such as the Medicare Shared Savings Program (MSSP) and ACO REACH ending this year. The shrinking list of competitors in this space include Aledade and Astrana Health (which bought one of the pioneers, Collaborative Health Systems).  Release, MedCityNews

Part 1: Midjourney Medical and the Prototype Scanner

Can the reality match the hype that’s jumped the rails? Midjourney Medical is a new company (or division) that grew out of Midjourney, a organization claiming to be a “community-funded lab of 60 people known for building the most beautiful AI models in the world”. Moving past the breathless prose, Midjourney Medical has developed a prototype of a whole-body scanner that, unlike present CT scanners and MRIs, does two major things: it fully scans the body in a minute and there is no radiation. The methodology is almost Star Trek-ian: your body is lowered gently into a circular tank of warm water, and 40 chips arrayed in a ring generate waves that change shape when they meet parts of your body (e.g. from skin to fat to muscle to bone).  The waves generated by the chips are translated into  ultrasound computed tomography (USCT) creating a 3D map of a human body, with the completion goal at 60 seconds of scanning and computing time.

This is coming from a 1st generation prototype. The images in the announcement video are supposedly not AI-generated but actual from about a dozen subjects. The actual timing is about 20 minutes at present, a real time not in the announcement but in articles (see below), and the subjects were uniformly slim.

Yet it’s audacious as all get-out, in a time drowning with incrementalism and no Big Next Thing.

If the results from the prototype are real and the timeline/vision are achievable, the potential is stunning. Being lowered briefly into a tank of water is a far more pleasant experience than the tube and noise of an MRI, though for claustrophobics or just those afraid of being submerged in water, the fears are similar. If taken to its logical development, it could, in fact, threaten the business of nearly all of the mainline MRI/CT scan companies such as GE Healthcare, Philips, Siemens, and Canon if it is medically validated and gains FDA clearance.

As Stuart Miller pointed out and described well in his article, there is already an FDA-cleared water/sensor array for USCT used as a diagnostic adjunct for dense breast mammography–SoftVue, developed by Delphinus Medical Technologies of Novi, Michigan. So there is precedent, one already in place and accepted in mainstream medical practice. A device that every woman would accept as standard instead of current.

However, breast scanning does not go through bone (e.g. ribs) nor gas (lungs), which is USCT’s drawback. At present, when ultrasound is used, a trained technician has to work the device between the ribs a millimeter at a time to gain an unblocked view, not in 60 seconds but taking 60 minutes or more.

Midjourney Health’s ambitions are wellness, not medical grade, at least into 2028. The announcement states that this prototype is the first of three.

• Their roadmap is that by end of 2027, Midjourney’s will open a 24/7 medical spa in San Francisco, with the 2nd generation prototypes providing detailed body composition maps for clients, accumulating data for its own case, scanner development, and yes, FDA.
• 2028 would introduce a 3rd generation scanner with custom chips and a ‘night-and-day’ change in image quality and scan times.
• The trajectory is opening proprietary medical spas, staying within ‘general wellness’ permitted by FDA, and gaining acceptance of the technology.
• Ultimately, by 2031, the plan is “to have a fleet of over 50,000 scanners worldwide – with a total scanning capacity of a billion scans a month”. The rationale: early imaging without symptoms>>change in lifestyle>>avoidance of 30% of deaths and 50% of healthcare costs.

Mr. Miller’s skepticism in his article is well presented and warranted. He details that ultrasound can’t go through bone and gas, dunking has psychological and physical drawbacks such as consistently clean water, the ability to run spas at scale, the business drawbacks of no reimbursement, no billing codes, no FDA validation. Unless it is priced well, the Midjourney vision of a pop-in spa and wellness checks  will be confined to the early adopters and the affluent curious privately paying up to thousands of dollars for the experience, much as whole-body MRIs, also in the ‘general wellness’ category away from FDA scrutiny, are today. Can spas support 50,000 scanners? Not likely.

It all sounds rather vague and glossy in a song that this Editor has heard before in the digital health (IBM Watson Health circa 2012) and concierge practice areas.

What is also scarce in this announcement is that once you have this information, what do you do with it? Is it conclusive enough to take to a physician and do follow up?  Or if lifestyle, will these spas affiliate with clinicians who can do the counseling? From the Forbes article on the announcement:

“Whole-body screening of people with no symptoms turns up incidental findings in 20% to 40% of scans, yet only a small fraction ever require treatment, as University of Michigan radiologists have noted. At a billion scans a month, even modest rates imply hundreds of millions of ambiguous results a year, each one demanding a clinical decision and producing a worried customer.”

In the category of ‘it does what it says it does”–a must in generating credibility for a new technology–false positives or even ambiguity are killers.

Since Midjourney Medical is a separate community-backed research lab, they are seeking support from the ‘community’ and not, at this point, explicitly seeking investors. There are reports that the company generates hundreds of millions of dollars in annual recurring revenue, according to Mobihealthnews. Their substantial funding needed for development, chips, and spas has to come from somewhere, and this vagueness about their funding disturbs this Editor.

What Midjourney Medical has in spades is strong skills in generating press coverage. Besides Stuart Miller’s Substack article on 19 June, Mobihealthnews published their take the same day, along with a wave of articles over the next ten days from Radiology Business to Fox News’ Cyberguy to Futurism. The Verge’s 23 June critique proved to be a lengthy deflation. While it presented Midjourney’s side by interviewing their head of medical Tom Calloway, the bulk of the article consisted of far more critical comments from several imaging and radiology professionals. Matthew Davenport, a professor of radiology at the University of Michigan Medical School, said that (FTA) published images were “interesting” and that he could see a market for body imaging. But the company’s “claims are wildly unsubstantiated, perhaps the most grandiose” he has seen. William Morrison, a radiology professor at Thomas Jefferson University, was downright scathing calling it a “vibe-based rollout” and (FTA) that the whole move, he said, has the feel of an ad campaign. “It makes me think that this may be more of a grift than a pivot.” Words like “grandiose” and “grift” tend to stick. 

Part 2: Butterfly Network, Their Chips, and the Agreement with Midjourney

Midjourney Medical had a partner in this which, interestingly, was not mentioned in the announcement. Our Readers may be familiar with the name Butterfly Network. It’s a company we covered extensively from 2018 when point-of-care ultrasound (POCUS) was being revolutionized by less expensive handheld units feeding into mobile software, powered by a new design of capacitive micro-machined ultrasound transducers (CMUTs) that are integrated into semiconductors. They are now up to their third generation iQ3 ultrasound device with improved imaging while continuing to market the second generation iQ+ handheld across multiple medical specialties and veterinary medicine. Their software to analyze images is a SaaS subscription model. Butterfly Networks also prominently offers on its website its software development kit (SDK) to developers and to providers, third-party software through its AI Marketplace. 

What they don’t publicize other than one website page, indexed in ‘About Us’, is Butterfly Embedded–selling the Ultrasound-on-Chip, the unique silicon that makes the iQ work. The Midjourney deal was revealed through a Form 8-K SEC filing published last November. Butterfly Network disclosed a Material Definitive Agreement with Midjourney, Inc., which gave them a non-transferable, limited license for the Ultrasound-on-Chip technology, software, and backend technology. From the filing: as part of a five-year agreement, Midjourney paid a one-time non-recurring fee of $15 million and a $10 million annual license fee plus (i) additional payments of up to $9 million upon the achievement of specified milestones, (ii) certain revenue sharing payments in connection with Midjourney’s commercialization of hardware products incorporating Company chips, and (iii) payments in connection with any purchases of chips from the Company.

This is no small change for Butterfly Network. It not only boosted their bottom line but also their long-languishing share price. BFLY went public in a 2021 SPAC that cracked hard by 2023, dropping in 18 months from a debut at $19/share to $2.42 and fell even lower to the dollar range where it stayed for most of 2024, with losses to match. It flirted hard with the Devil of Demise that claimed many SPAC’d companies, yet survived under changed management, despite a roller-coaster 2025-early 2026.

With the Midjourney announcement, the stock went up to the $8 range, double from May. This was aided by Butterfly’s own announcement of the deal, interestingly as “commentary” on Midjourney’s announcement. It positioned it as a collaboration enabled by Butterfly Embedded, Ultrasound-on-Chip licensing, and a co-development business initiative. The release preceded an Emerging Growth Webinar with TD Cowen, which was fortuitous timing indeed.

A week later, Butterfly announced a second Embedded deal with Aleph Neuro for what they termed the “highest-resolution 3D images of the human brain taken from outside of the skull” without drilling or MRI. Aleph is a pure research lab and did not present a growth business case. The only two previous chip/Embedded deals featured on their website dated back to 2023: Mendaera and Forest Neurotech.

Pulling On The Loose Threads

My comment on Stuart Miller’s first Substack was around Butterfly’s ability to scale production of the chip for the prototypes and the projected 2027 rollout. I also noted that Butterfly never took off as we in digital health had expected. It had strong press, financial backing, and a unique technology, yet mainline companies with piezoelectric ultrasound didn’t jump to buy Butterfly when they could have for a tuck-in price, nor did providers jump to adopt it.

His second article, The Med Tech Company Behind the Midjourney Scanner, pulled on my ‘loose thread’ and showed why Butterfly ultrasound handhelds, despite their advanced technology, didn’t have a chance–how acceptance by hospitals and clinicians was a long uphill slog that an upstart company could not win against GE Healthcare, Philips, Siemens, etc. They all now have their own handhelds to mobiles. As I put it in another comment, no one gets fired for buying any of those three. Butterfly’s POCUS moat, dug deep on CMUT, has filled in to be shallow and narrow as far as providers are concerned.

With the Aleph and Midjourney deals, Butterfly with Embedded has now pivoted to the very, very sexy chip space. A luster that may help out their main POCUS business but functionally gives them an entirely new revenue stream. The company broke over the past week a heavy ad schedule on X and LinkedIn for the new iQ3, now on sale and with some attractive discounts. They’re taking advantage of strong positive news and being seen in a different light, their unique chip covered with a golden glow at last. Butterfly is also being touted by various stock pickers as a hot, undervalued buy, with estimated earnings from the Midjourney deal perhaps $74 million over five years. Here’s an example on X from CK Capital. 

Instead, what should grab your attention are two short sentences in the Midjourney announcement (that doesn’t mention Butterfly) that Mr. Miller points out:

“In 2028, we’ll start scaling to more cities, and we’ll upgrade to a 3rd generation scanner. Gen3 is where it gets ‘serious,’ the silicon for this design will be completely custom, and image quality and scan times will be night-and-day.”

2028 is within the five-year agreement. Yet he poses certain logical questions that should be plain to any investor.

• Are the Butterfly chips only a bridge until Midjourney finds or develops something better on its own?
• Suppose Midjourney finds a more powerful or connected chip partner?
• What happens if Butterfly can’t scale production for the 2nd generation?
• Is the deal that Butterfly will be the custom chip designer–or not?
• And the killer-diller–Will the Midjourney scanner even make it to Gen3?

Will Butterfly be able to leverage these (at long last) good times into increased POCUS sales–more importantly, a buyout or a go-private with fresh investors? Stay tuned.

This week’s Big Data Breach affects 1.4 million patients at multiple healthcare organizations.

The multi-national data security alliance Five Eyes warns of AI supercharging hacking attacks. The three-page statement details how AI accelerates cybersec attacks and the need for ‘defence in depth’ with threats increasing in months, not years. It offers a five-point plan to reduce vulnerabilities and to use AI to defend against attackers. However, with most healthcare organizations overwhelmed with implementing AI tools, suppliers like Xsolis a vector for attack, and employees going outside for AI tools, the threat level has been amped 100x. The MIT Sloan article also warns that Anthropic’s Mythos, which is reportedly capable of autonomously finding and exploiting software vulnerabilities end-to-end with no human involvement, could be used for cyberattacks and chemical/biological attacks. Five Eyes is drawn from the US, UK, Canada, New Zealand, and Australia cybersecurity agencies. DataBreaches.net

In more cheerful news, Whoop announced that the FDA closed an investigation into their wearable’s Blood Pressure Insights feature. This started with a letter from FDA’s Center for Devices and Radiological Health (CDRH) in July 2025 challenging Whoop on the basis that the company did not have an approved application for premarket approval (PMA) or 510(k) approval of that feature.  Apparently, Whoop backed off of original claims that the BPI offered medical-grade health and performance insights and moved to general wellness claims ‘not intended for medical use’. It didn’t dissuade funders from a gigantic $575 million Series G in April. [TTA 9 Apr] MassDevice 

Centene, which in last week’s Chutes announced a voluntary separation plan (VSP) to most employees with 2 years or more in the company, added a board member. Lauren Tyler will be joining immediately to serve on Centene’s audit committee and compensation and talent committee. Ms. Tyler is a 20 year-plus JP Morgan veteran who was global head of human resources for asset and wealth management, global firmwide chief auditor, and global head of investor relations. That is an interesting skill set given what is happening at Centene and the need to compensate by downsizing for the crash in Medicaid and ACA members, as noted in the Release, Healthcare Dive

Raises have perked up again after a few weeks off.

Assort Health just raised a $120 million Series C. It was led by Menlo Ventures, with participants including Lightspeed Venture Partners, Felicis, First Round Capital, Chemistry, Joe Montana (!), Tau Ventures, and Quiet Capital. Assort provides an AI-assisted patient  voice agent along with an impressive AI model, Synapse, for specialty workflows that automate scheduling, intake forms, referrals, document processing, medication refills, real-time eligibility, lab requests, and payments. San Francisco-based Assort to date has raised $222 million and is now valued at $1.2 billion. Release

xCures scored a $46 million Series B. It was led by Innovius Capital, with participation from iGrow, GKCC, Spring Mountain Capital, and existing investors. Total funding is now over $76 million. xCures’ business model is focused on gathering scattered clinical patient data, assembling and structuring patient medical records into usable decision-ready data through its Clinical Clarity Engine. It is delivered to users via a web UI or a developer-friendly API. To date, they have 300 million medical records sourced from more than 550,000 healthcare locations nationwide. The new funds will be used for expansion of the Clinical Clarity Engine’s capabilities. Release, Mobihealthnews

The always controversial Ed Zitron goes there. Way out there to challenge your preconceptions. If you’ve ever awakened at 3am wondering where in blue blazes AI is going, agonizing over where it fits in your platform or services, who is it really benefiting other than OpenAI and Anthropic, why VCs are so stingy because your tech isn’t AI All Over, and why despite the hype billions are going down a rabbit hole seemingly with no bottom [TTA 2 June, 25 June] –this article (not premium paywalled) is for you.

“Cargo Culture” refers to the cargo cults of the South Pacific. They originated after native contact with Westerners, usually Americans who landed in ships or built airstrips, mostly during WWII. When the Americans left, the indigenous peoples tried to recreate the “magic” of goods coming off the ship or from airplanes by creating a religion and ritual around these ‘miracles’ without, of course, the corresponding methods of delivery. Your Editor also found an IT version of this called “cargo cult programming“defined in a Wikipedia entry as “computing slang to describe the inclusion of code that serves no purpose in a program, indicating a lack of understanding of the program structure by the programmer.“

Mr. Zitron’s “Cargo Culture” refers to Silicon Valley. The endlessly reinforced belief system of the interlocked software, venture capital, and private equity businesses was built on decades of mythology of endless growth and endless Christensenian “disruption”. They want to believe that there will always be hypergrowth tech companies, always another ‘disruptor’. And the money to fund them will be equally endless. (Forgotten are the failures like Theranos and the billions down the drain.)

The reality, as he painstakingly explains, is that Silicon Valley and tech are, right now, fresh out of ideas. Worse, the software industry is declining. There’s no Google Search or iPhone waiting to be plucked from the ether. There is no New Big Transformative Idea. Company growth is incremental at best, declining or failing at worst. But instead of adjusting, the mythology is ingrained, and to doubt is heretical.

Any dissension, whether corporate or media, means you are Voted Off The Island.

The AI Bubble fits neatly and virally into Silicon Valley’s Cargo Culture and need for a disruptor. It’s a viral product, it was different, it meets a need for automating processes. It makes search a lot quicker. But will LLMs and agentic AI actually change anything other than make it faster, or reduce the need to perform your own logic in building a case/article–at amazingly high expense? Other than the spend, what’s exponential about it? Everyone’s buying a lot of GPUs–heck, Oracle at Larry Ellison’s behest is betting the company on moving out of software, including that ever-so-sticky EHR–and becoming a data center constructor and lessor to Open AI and Meta lest Microsoft eat his lunch. After a certain point, the beliefs become circular and self-reinforcing. 

Mr. Zitron argues the points at length. It’s minimum a 30-minute read and not an easy one. Stay with it. It challenges a lot of premises. It will make you think differently the next time you’re pitched on a company’s miraculous AI software–or if you are a developer, understanding your customer’s needs and likely confusion. It is 100% guaranteed to make you think about Silicon Valley, their modus operandi, and how it’s inflated the whole AI bubble.

Sergei Polevikov shines a light on a recent Nature Medicine study comparing specialized clinical LLMs against the generics–and finds that the generics win. The peer-reviewed study pitted specialized clinical tools like OpenEvidence and Wolters Kluwer’s UpToDate ExpertAI against general-purpose AI tools–ChatGPT-5.2, Gemini 3.1 Pro, and Claude Opus 4.6–using three different benchmarks*. The specialized tools lost every time. What was worse was OpenEvidence’s unprofessional overreaction on social media, namely X. Instead of responding directly to Nature Medicine as a comment, then publishing a peer-reviewed paper refuting the Nature Medicine team findings, citing their own research from independent studies, conducting fresh research, even better admitting that some of the research team’s findings were valid and appreciated, their ‘Twitterstorm’ only created more PR blowback. Social media amplified the damaging reach of one academic study. (Having managed a few ‘crisis communications’ as a marketer, the worst thing that you can do is bang back in anger, though cooling corporate tempers might require a Titanic-sized iceberg of ‘splainin’.)

Mr. Polevikov’s free excerpt of his lengthy article, OpenEvidence Goes Hippocratic AI, is published on LinkedIn, but why not kick in a few $$, support his independent journalism, and subscribe to AI Health Uncut (Substack)?

The Hippocratic AI reference is to that company’s ‘bare knucks’ approach to any criticism of their platforms. 

*FTA Nature Medicine: (1) 500 MedQA questions testing medical knowledge, (2) 500 HealthBench items measuring alignment with clinicians and (3) the real clinical queries (RCQ) benchmark, built from 100 de-identified queries from physicians to a general-purpose language model in a live clinical environment. 

Bain and Company dug into the Big Question surrounding AI–is there any real ROI there? How can companies get there?

• Nearly 40% of companies that measured AI cost savings landed below 10%. Another 29% achieved 11-20%. Neither are huge savings.
• Yet 90% of companies are increasing their AI budgets, again. How they’re funding it? 44% of companies are funding it from prior automation savings that have consistently come in below target.
• The operating reality is far more human, with only 7% of companies running fully autonomous agents in production today.
• Data access and integration are the number one barriers to AI progress

Despite all this, Bain (of course) states that companies shouldn’t wait, that they need to close the gap between the business case and the operating reality. Their conclusions sound sensible, but to this Editor, they won’t necessarily bend the cost savings curve. And they sound like expensive, time-consuming work, further reducing initial cost savings:

1. Pay down your workflow debt before deploying AI. Fix that before you automate it because it will be far more difficult afterwards.
2. Validate the investment case and name a governance owner before programs launch.
3. Use AI to solve the data problem; don’t wait for the data problem to be solved first. Automate one high-value, repeatable workflow and replace it with AI.
4. Redesign the operating model, not just the process.
5. Measure outcomes at the enterprise level, not the program level. Leaders must decide that they have a personal responsibility to create the organizational conditions for AI success. 

Your AI Budget Is Growing. Your Returns Aren’t. Here’s Why.

Our second report is on innovation in the women’s health sector. Silicon Valley Bank (SVB) examines trends in the venture-backed women’s health sector, which has grown into the full spectrum of women’s health needs. The sector is (finally) starting to thrive. Highlights:

• $2 billion in venture capital was invested in women’s health in 2025. While this marked a dip from 2024’s peak, projections show a rebound in both deal volume and total funding for 2026.
• $6.2 billion since 2019 has been invested into companies addressing women-specific conditions. 210 new startups dedicated to solving women-specific health challenges were launched from 2019 through 2025. 
• The median pre-money valuation for AI-enabled women’s health companies is up to $35 million.
• Women’s health startups using AI to personalize care are commanding valuations triple those of their non-AI counterparts.

The downloadable SVB report is here.

One of those women-founded companies, Ladder Health. announced seed funding of $7 million this week. The seed round was (happily) oversubscribed and led by Nina Capital, with participation from Mairs & Power Venture Capital, South Dakota First Capital, and incubating partner 25madison Health. Other investors in this round include Hatteras Venture Partners, Create Health Ventures, Jumpstart Capital, White Oak Enterprises, Groove Capital, and 7Rock Ventures. (Whew!) Ladder Health delivers speech, occupational, physical and feeding therapy through a virtual-first, AI-enabled platform available evenings and weekends. Their service is targeted to areas where pediatricians and developmental therapies are in short supply, wait lists are long, and to underserved Medicaid beneficiaries. It is currently available in three states, North Carolina, Massachusetts, and Maryland across 80 providers. The additional financing will support expansion in those states and others, as well as investment in the care platform. Ladder is based in NYC. It was developed in conjunction with clinical experts at Boston Children’s Hospital. Release, MedCity News

And lest we forget. Luigi Mangione, the accused murderer of UnitedHealthcare CEO Brian Thompson on the dark morning of 4 December 2024, is still in the preliminaries of his New York State trial for intentional second-degree murder. His defense team from Agnifilo Intrater had filed to use a psychiatric defense based on Mangione suffering from an “extreme emotional disturbance” (EED), but that was withdrawn one day after it had been filed with Manhattan Supreme Court Judge Gregory Carro. It was a surprising development, as the admission could have reduced the charges to manslaughter, though it is tantamount to admitting the crime.

Based on reports, Mangione may be able to introduce EED through his own testimony, or that of the prosecution. Much of the evidence found in his backpack when Mangione was picked up at the Altoona, Pennsylvania, McDonald’s was excluded from the NYS trial, as well as two charges of terrorism and first-degree murder. He has pleaded not guilty to state murder and weapons charges.

Mangione’s trial is scheduled to start on 8 September. He faces additional trials in the Federal Southern District of NY as well as Pennsylvania state charges. Law.com discusses the EED strategy. CNN, Beckers

Pepper faints again, reading Amazon’s response–it’s ‘archived’ data that “only impacts certain legacy Iora Health and One Medical Seniors patients”. 

Amazon One Medical published a notification on its One Medical Seniors (the former Iora Medical) website summing up the following:

• They learned on 13 June that a third-party file storage system used to store archived patient information had been accessed by an unauthorized person
• Data had been exfiltrated between 8-11 June
• It’s apparently “legacy” information only and affecting only certain old Iora/Amazon One Medical Seniors records
• The clinics affected are legacy Iora clinics located in Atlanta, Cape Cod, Charlotte, Piedmont Triad, Denver, Houston, Phoenix, Tucson, and Seattle.
• They took immediate action including the revocation of all user access and rotating credentials for all employees with access to the system. 
• Neither Amazon One Medical (non-Seniors) patients, Seniors patients outside these markets, nor their EHR and medical services have been affected.

Patients are given a number to call and an email to write if they have questions. It is not known if patients were notified in other ways, for example through the patient portal, email, or standard US Mail. There is no public indication that Amazon One Medical has contacted ShinyHunters or is negotiating with them, whoever they are.

ShinyHunters is notorious for taking credit for other attacks on healthcare organizations, always going big, then going home after gaining their ransom. Their last target was Medtronic, hacking 9 million patient records plus terabytes of corporate information in April [TTA 30 April]. After the leak information was pulled from their website, the conclusion was that Medtronic paid up. ShinyHunters also attacked dental benefits administrator DentaQuest earlier this year. Other big game: Zara, Carnival, 7-Eleven, Pitney Bowes, The Canada Life Assurance Company, and Hallmark. 

Amazon One Medical is staying mum, but this will be updated as additional information is disclosed. Cybernews, Healthcare IT News, HIPAA Journal

The doldrums of M&A and fundings lightened a bit this week.

Personal health data consolidator Validic acquired by ChartSpan. The care management services company will broaden its portfolio through integrating Validic’s IoT platform into its remote care management and health programs. These include chronic care management, advanced primary care management, remote patient monitoring, and custom programs for health systems, payers, and wellness programs. Acquisition cost was not disclosed. BIP Capital led the financing for ChartSpan, as they did with their $15 million Series A in 2019. Validic’s last raise was $12 million in 2022 led by Kaiser Permanente Ventures. Validic, founded in 2010 and still led by founder Drew Schiller, will be integrated into ChartSpan. No mention of how the workforces will be integrated. Validic is headquartered in Durham, North Carolina, ChartSpan in Greenville, South Carolina. This continues the consolidation of complementary health tech businesses.   ChartSpan blog, Mobihealthnews

Short takes on recent raises:

Cadence Solutions, another clinical intelligence and services company in chronic care management, raised $100 million in a Series C. The round was led by Spark Capital. Other participants were Thrive Capital, General Catalyst, Coatue, B Capital, Corewell Health Ventures, Memorial Hermann, and Duke Health. Cadence’s total funding since 2021 is $241 million. In 2021, they were valued at $1 billion though their current valuation is not public. The fresh funding will be used to expand across new health systems, advance Cadence’s AI agents, and grow value-based care models. Cadence also announced new affiliations with Duke Health and Texas Health Resources, now serving 20 leading health systems including Memorial Hermann in Houston and Hartford HealthCare. The model is a little different than similar remote patient monitoring (RPM) services. These health systems ‘white label’ Cadence’s clinical teams, who follow the health systems’ protocols in monitoring the device information and adjusting care in real time. Cadence now serves more than 100,000 active patients. Cadence release, Mobihealthnews, MedCity News

Prosper AI gained a $30 million Series A. Funding was led by Andreessen Horowitz/a16z, with participation from Base10 and continued support from Emergence Capital, Y Combinator, and Company Ventures. Prosper AI developed a series of voice agents for administrative workflow tasks such as answering patient calls, scheduling appointments directly in the EHR, verifying insurance benefits, automating patient billing, and contacting insurers on the phone when additional information is needed. The breadth of the agentic AI claims to lower administrative costs by +40% and up to 50% of end-to-end patient conversations. The company claims 150,000 healthcare providers and has added more than 40 care organizations as customers since their last funding round in September 2025.  The new funds will be used to expand its engineering and customer-facing teams, deepen integrations across the largest EHR platforms, and accelerate adoption across provider groups and health systems. Prosper release, Mobihealthnews

Brazil’s Telepatia raised a $33 million Series A, mainly from Andreessen Horowitz/a16z. Additional participants were Palantir CTO Shyam Sankar, Rappi founder Simón Borrero, and Nubank founder David Vélez. Telepatia combines AI-powered documentation, clinical decision support, and healthcare assistants to help clinicians improve productivity, reduce medical errors, and increase adherence to clinical protocols. It is targeted to healthcare systems in Latin America that have shortages of physicians and nurses. Telepatia has already been adopted by 25 hospital systems in Brazil, Colombia, and Mexico, reaching 14 million patients, improving protocol adherence from 84% to 99%, and helping prevent 60,000 medical errors in real time,” according to a16z’s announcement post. Mobihealthnews, The SaaS News

Patients are using portals for messaging and other tasks, but using them along with increased physical visits. A five-year study by NYU Langone Health researchers of the Epic MyChart system, published this week in the Journal of the American Medical Association (JAMA), found that:

• 12% of Americans now use patient portals for messaging about appointments, test results, and ongoing treatments
• Online portal messages more than doubled between 2020 and 2025 (153%)
• Telephone calls decreased by 6%
• Americans with an active Epic health record went from 94 million in 2020 to 140 million in 2025.
• 30% of the 42 million active patients on Epic sent a portal health app message to their clinician during the first three months of 2025.
• Individual patient messages doubled, moving from an average of 2.2 per year in early 2020 to 5.4 per year in late 2025.
• In-office visits returned to an average of between two and three per year per patient

The study is the largest review conducted of Epic EHR records: more than 140 million patient records from 2,067 hospitals and 47,100 health clinics in the US. NYU Langone release