Making ChatGPT better for clinicians
OpenAI introduces ChatGPT for Clinicians, which offers advanced healthcare-specific models and allows users to create reusable skills such as writing referral letters and searching peer-reviewed journals.
AI-driven RCM services provider IKS Health will acquire TruBridge for up to $565 million.
Almanac Health, which offers AI clinical decision support, raises $10 million in seed funding.
OpenAI introduces ChatGPT for Clinicians, which offers advanced healthcare-specific models and allows users to create reusable skills such as writing referral letters and searching peer-reviewed journals.
ChatGPT for Clinicians is offered at no charge to NPI-verified US clinicians. It also offers CME for asking clinical questions.
The product is designed to support a single clinician’s workflow as an assistant, as opposed to the enterprise-focused ChatGPT for Healthcare, which offers more compliance and governance capabilities.
The announcement suggests that companies such as OpenEvidence are seeing their core functionality commoditized by a globally familiar platform in a market where clinicians are likely to standardize on a single tool. Differentiation will hinge on transparency and trust, the quality of the information sources that are used, the capability to summarize information accurately, and the depth of integration with existing systems.
From Bill Coed: “Re: No Surprises Act. Regulators assumed that it would protect patients and that doctors would play fair. Instead, it’s a lucrative workaround for providers who are willing to work the arbitration system in return for massive rewards. As always, financial incentives affect behavior more than the rules themselves.”
I needed a 256GB MicroSD card and learned that memory prices have spiked as AI demand has consumed manufacturing capacity. I expected to pay about $20, but the best option I found was a SanDisk ImageMate Pro for $60 at Walmart, since Best Buy, Office Depot, and others were out of stock on brand name cards. Pro tip: don’t buy memory cards or other electronics from Amazon since third-party counterfeits are common and even Amazon-fulfilled inventory can be contaminated by returns. Not to mention that Amazon’s hardening line on returns makes it difficult to get credit if you order a laptop or other item and instead instead receive some random object from a scam seller that weighs the same, which is hard to prove unless you video your unboxing. I’m actually thinking of dropping Prime and moving to Walmart+ since I’ve had great success with “shipping” that often arrives the same day and even with Uber-like driver tracking and messaging, at least from my experience yesterday.
None scheduled soon. Contact Lorre to have your resource listed.
AI-driven RCM services provider IKS Health will acquire TruBridge for up to $565 million.
Almanac Health, which offers AI clinical decision support, raises $10 million in seed funding.
Med Tech Solutions appoints Dan Stoke (CTG) as chief growth officer and Matt Trevorrow (NTT Data) as CIO.
Inbox Health hires David Henderson (GroundGame.Health) as CFO.
Michael Han, MD, MBA (MultiCare Health System) joins Ambience Healthcare as chief medical officer.
Sandhills Medical promotes Amanda Duke to CEO.
Arrowhead Regional Medical Center appoints Andrea Daugherty, MHA (SHI International Corp.) as chief information and digital transformation officer.
A Nordic survey finds that managed services in healthcare IT have shifted from temporary staffing support to a core strategic function, with most health systems using them to drive long-term performance, modernization, and operational capacity.
A New York Times report finds that physicians and provider groups are using the No Surprises Act to increase payments by routing out-of-network charges into arbitration, where arbitrators often award amounts that exceed typical insurer-paid rates. The federal government expected 17,000 cases to go to arbitration each year, but doctors filed 1.2 million cases in the first half of last year alone, winning 88% of them and generating $885 million in arbitrator fees. Arbitrators must choose either the insurer’s offer or the provider’s requested amount, and their decision cannot be appealed. A neurosurgery practice took a $2,700 diagnostic procedure to an arbitrator who awarded it $333,000. A plastic surgeon who charges $15,000 to $25,000 for breast reduction surgery has earned $440,000 for the procedure via arbitration. Health plans are raising their premiums to cover the extra costs.
Blog Posts
Mr. H, Lorre, Jenn, Dr. Jayne.
Get HIStalk updates.
Send news or rumors.
Follow on X, Bluesky, and LinkedIn.
Sponsorship information.
Contact us.
It’s all AI this week.
An article in the European Heart Journal examines whether AI analysis of mammography can go beyond identifying breast cancer to also predict the risk of heart disease. The study involved 123,000 women who underwent imaging at the Mayo Clinic and Emory University. Patients were followed for a median of seven years to look at whether events such as heart attack, heart failure, and stroke occurred. The authors found a correlation between calcifications in breast arteries and major cardiovascular events. Additional research is needed to reproduce the results across different scanner manufacturers and patient populations, but this could be a promising way to screen women for cardiac risk without additional radiation exposure or time away from work or home.
Several new articles looked at the accuracy of AI tools. One that was published in BMJ Open looked at how general purpose AI tools handle the kinds of questions that patients might ask about health topics that are prone to misinformation, such as cancer, stem cells, vaccines, nutrition, and athletic performance.
The authors used 10 prompts in each category to test Gemini, DeepSeek, Meta AI, ChatGPT, and Grok. They used a “red teaming” strategy to try to provoke misinformation and had two experts assess each response and score them on how problematic they were based on predefined criteria. They also scored the models for completeness, accuracy, and readability.
About half of responses were identified as problematic, with nearly 20% being highly problematic. Grok generated more highly problematic scores than expected under a random distribution. Gemini had the fewest problematic responses.
The paper notes that responses were “consistently expressed with confidence and certainty,” and that out of 250 questions, only two resulted in failure to answer. Both of those were from Meta AI.
Readability was also challenging, with responses scoring at a college reading level.
One limitation of the study was that the results were not blinded. The people doing the scoring knew which models had produced the answers that they were evaluating.
The authors note that AI tools are rapidly evolving and that their findings might not be generalizable to other tools or to paid versions of the free tools that they studied.
The included questions whether 5G technology or antiperspirants cause cancer, whether mRNA vaccines alter the human genome, the amount of raw milk that is needed for health benefits, and whether anabolic steroids are safe. The article highlights hallucinations and fabrications in the requested references:
Models appear to recognize their referencing limitations but are unable to address them. In one of our interactions with DeepSeek, the model acknowledged that its references were generated from patterns in training data “and may not correspond to actual, verifiable sources.” It also noted that ‘Synthesized references (e.g, author names, publication dates, or titles) might be approximate, outdated or occasionally fictional.” When we asked ChatGPT-3.5 to explain its poor citation reliability, it responded: “ChatGPT may fabricate information to maintain the appearance of completeness — even if that means sacrificing accuracy.” Our data support the contention that the current generation of chatbots may not be suitable for tasks requiring a high degree of factual accuracy or verifiable sources.
This study was designed to approximate a patient-driven conversation. I wonder how many patients are aware of the degree to which various models simply make things up? Early in my career when patients began embracing the internet as a source of medical information, we had conversations about which sources were reputable and which might not be the best. At one point, I had a patient-facing handout that listed my favorite reputable sources. I wonder how many organizations might be creating these kinds of educational materials for patients to help navigate the world of AI-generated medical advice.
A clinician-focused article in JAMA Network Open evaluates whether off-the-shelf large language models (LLMs) can perform reliably in clinical situations. The study’s objective was “to evaluate the longitudinal clinical reasoning ability of state-of-the-art LLMs and to introduce a multidimensional, clinically meaningful benchmark for clinical-grade artificial intelligence (AI).” They identified 21 LLMs to evaluate, including GPT-5, Claude 4.5 Opus, Gemini 3.0 Flash and Pro, and Grok 4.
The created 29 clinical vignettes and had medical students score the performance of the models across clinical reasoning: differential diagnosis, diagnostic testing, final diagnosis, management, and miscellaneous clinical reasoning questions. The clinical vignettes included basic patient demographics such as age and sex as well as a history of present illness, review of systems, physical examination findings, and laboratory results.
The authors found that the models performed well at answering diagnosis and management questions, but performed worst when asked to formulate a differential diagnosis. The authors concluded that current models are limited in their ability to conduct diagnostic reasoning and that the models should not be used for unsupervised patient-facing clinical decision-making.
One limitation of the work is that the researchers followed a stepwise process in presenting the vignettes through question-and-answer format, which might not be how clinicians use these tools in the real world. The vignettes were also publicly available, so there are no guarantees that the model didn’t have access to them prior to the experiment. The assessment also didn’t include additional features such as calculators, agentic tools, access to guidelines, or retrieval-augmented generation. The authors noted that this means “the results reflect baseline longitudinal clinical reasoning rather than maximal achievable performance.”
They go on to say that “the findings of this study caution against vendor claims that general purpose, off-the-shelf LLMs are ready for patient-facing clinical use.” Based on my use of several of the models as I complete my continuing medical education coursework, I would agree. I’ve seen enough questionable answers to remain wary, and often wonder how well clinicians with less experience can identify the answers as problematic or whether they just take them for fact. The authors suggest that the “most responsible role” for off-the-shelf LLMs at this time is “targeted, clinician-supervised use in low-uncertainty tasks.”
Is your institution using commercial LLMs for patient care? Have they restricted access to such tools via network controls, forcing clinicians to furtively access them on their phones? Leave a comment or email me.
Email Dr. Jayne.
Amperos Health, which offers an AI-based denial management and revenue recovery platform, raises $16 million in a Series A funding round.
Kipu Health Acquires Team Recovery Technologies, Its Largest Deal Yet
Behavioral healthcare software company Kipu Health acquires Team Recovery Technologies, which specializes in patient and alumni engagement and consulting services for behavioral health treatment programs.
AcuityMD raises $80m for AI augmentation to medtech sales platform
AcuityMD, which offers medical device sales software, raises $80 million in a Series C funding round.
Michael and Susan Dell donate $750 million to the University of Texas at Austin to fund a new AI-native medical center and research campus. Michael Dell says that building from scratch allows data, computing, and AI to be embedded from the start, enabling earlier decisions, better care coordination, and improved outcomes. He left UT Austin as a pre-med sophomore to start Dell Technologies.
Meta Platforms will install tracking software on the computers of its US employees to capture their mouse and keyboard activity, along with screenshots, for AI training. The company says that it needs the data to help its models understand real-world usage and promises that it won’t use the information to review job performance.
AI chatbot apps are marketing themselves as providing “therapy” while disclosing in the fine print that they do not offer advice, diagnosis, or treatment that should be provided by a licensed professional. Experts warn of a lack of evidence of effectiveness, the absence of FDA standards, and the tendency of chatbots to empathize or flatter users instead of redirecting conversations toward issues the user is avoiding.
Smart ring maker Oura acquires Galen AI, a year-old startup that offers an AI-powered personal health companion.
Denial management technology vendor Amperos Health raises $16 million in a Series A funding round.
AcuitMD, which offers medical device sales software, raises $80 million in a Series C funding round.
Specialty medical practice software vendor ModMed acquires Bonsai Health, which automates front-office workflows with agentic AI.
In Australia, whistleblowers at Royal Darwin Hospital say severe understaffing, poor hygiene, and inadequate training are compromising patient care, with nurses sometimes relying on YouTube for training and ChatGPT to calculate medication doses.
An elite Wall Street law firm that bills $2,000 per hour admits in an $8 billion bankruptcy case that rival firms correctly identified AI hallucinations in its emergency filing, including nonexistent citations and misquoted laws. The firm, which advises OpenAI on AI ethics, says that its internal AI policies were not followed.
Mr. H, Lorre, Jenn, Dr. Jayne.
Get HIStalk updates.
Send news or rumors.
Follow on X, Bluesky, and LinkedIn.
Sponsorship information.
Contact us.
Two Curves, One Hospital Server Room: Why On-Prem AI in Healthcare Is Inevitable
By Thomas Kantecki
Thomas Kantecki is a health informatics student at the University of Central Florida.
Model efficiency and GPU power are compounding quickly. Where they intersect, sending protected health information to somebody else’s data center stops being the smart tradeoff that it was in 2020.
I think health IT is headed for a reversal that most people who are running clinical infrastructure have not fully priced in yet. My prediction is that within the next five to 10 years, a meaningful share of hospitals and health systems will run their clinical AI workloads on hardware that they own, inside their own networks, on data that never touches a WAN link.
The driver is not a regulatory panic and not a breach cycle. It is math. Two curves are bending at the same time, and where they meet, sending protected health information to external data centers stops being the smart tradeoff that it was in 2020.
The First Curve: Models Are Becoming More Efficient
Five years ago, the idea that a hospital could run a capable large language model on a single workstation would have been laughable. Today it is routine.
The reason is not that hospitals bought bigger computers. It is that the models got smaller for the same quality.
A few things drove this. Mixture-of-experts architectures mean a 400B-parameter model can run with only 20B or 30B active at any given token, collapsing the memory and compute cost of inference. Quantization techniques like Q4 and FP8 cut the memory footprint of a model by 2x to 4x with minimal quality loss. Distillation has made smaller models remarkably competitive with their larger teachers on the specific tasks hospitals actually use them for: summarization, structured extraction, ambient documentation, code assistance.
The RTX PRO 6000 Blackwell, a single desktop GPU, can run a 70B-parameter model at FP8 with room left over for the KV cache. Five years ago you needed a small cluster to do that. The curve does not show signs of flattening. The same techniques that gave us 70B on a desktop are now giving us 400B on a server with better clinical performance than the 175B frontier of 2023.
The Second Curve: GPU Power Is Compounding Fast
Hardware is advancing just as quickly. Blackwell delivered roughly 2.5x better throughput per dollar than the prior Hopper generation, with inference costs dropping by up to 10x for workloads that can exploit the architecture. Nvidia is not slowing down, AMD is closing in on memory-bandwidth parity for inference, and purpose-built inference silicon from Groq and Cerebras has shown that the token-per-second ceiling has not been reached.
For hospital procurement, a capital purchase in 2026 to run inference on site will be outperformed by a capital purchase made in 2028, which will be outperformed by one made in 2030. The depreciation curve is real. It is also the same curve that makes each successive refresh cycle dramatically more capable for the same rack space and power budget.
Where the Curves Cross
The interesting moment is not today. The interesting moment is what happens when these two curves meet somewhere around 2028 or 2029.
At that point, a midsized hospital will be able to run what is effectively a frontier-class model, fine-tuned on its own de-identified corpus, on a single rack of GPUs sitting in a room the compliance team already has physical control over.
The performance difference between that local model and a cloud-hosted frontier model will be small enough that it stops mattering for the clinical tasks that actually touch PHI, such as drafting discharge summaries, ambient scribing, chart synthesis, and structured data extraction from notes.
At that point, the calculus changes. Today, CIOs ask whether hospitals can afford to run this on-prem. The question they will ask in 2029 will be why the hospital would send this off-prem at all.
Why Privacy Starts to Win the Argument
HIPAA Business Associate Agreements are legal instruments. They allocate liability and create enforceability. They do not change the physical reality that PHI is leaving the facility, traversing the public internet, and being processed on hardware that the hospital does not control, shared with tenants that the hospital has never met.
The expected 2026 HIPAA Security Rule removes the addressable designation for encryption and tightens requirements around asset inventories, network maps, and breach timelines. Those requirements are easier to satisfy when the GPU doing the inference is in a rack the hospital owns.
The privacy argument is not new. What changes is that the performance penalty disappears. Today, running only a local model means accepting a real capability gap versus the frontier. In three to five years, that gap will be small enough that the privacy side of the ledger dominates.
This Is an Opinion, Not a Forecast
I want to be honest about what this is. This is a prediction based on two trend lines and a belief that, given the choice, compliance officers and patients both prefer the version where the data never leaves the building. I could be wrong. Cloud vendors will keep improving their HIPAA posture. Edge-inference startups might hit supply chain walls. Regulatory bodies might carve out cloud-friendly exceptions.
The underlying trajectory is hard to argue with, though. Models keep getting more capable per parameter. GPUs keep getting more capable per watt. Patient data keeps being the single most sensitive dataset in the entire economy. Put those three facts together and the eventual destination is obvious. Clinical AI inference belongs inside the hospital. It just does not belong there yet.
The server room is coming back. The only open question is when.
ECRI Spins Out Healthcare Spend Management and Recall Management Solutions
Healthcare quality and safety non-profit ECRI spins off Staritas, which offers supply chain spend management and recall management solutions.
Michael and Susan Dell to donate $750 million to UT Austin to fund new medical campus
UT Austin will use a $750 million donation to build a research campus and medical center that will include a new hospital with AI embedded throughout its systems
Tava Health Lands $40M Series C to Boost 3 New Services
Digital mental healthcare company Tava Health announces $40 million in Series C funding.
Amazon launches GLP-1 weight loss program, promising ‘fast, convenient’ access
Shares of several competitors fall on the news that Amazon’s hybrid One Medical primary care business has launched a GLP-1 program with Amazon Pharmacy.
House FY27 VA funding bill allocates $3.4B for EHR rollout
The House Appropriations Committee proposes allocating $3.4 billion for the VA’s EHR modernization program as a part of the fiscal year 2027 funding package.
Healthcare quality and safety non-profit ECRI spins off Staritas, which offers supply chain spend management and recall management solutions.
ECRI has provided these products for 50 years.
Results from my side-by-side testing of ChatGPT and Claude, along with occasional frustration when ChatGPT wasn’t responding, have led me to pay $200 for a year-long subscription to Claude. So far, so good.
None scheduled soon. Contact Lorre to have your resource listed.
Ethermed, a Philadelphia-based prior authorization automation startup, secures $8.5 million in Series A funding.
Revenue intelligence software startup Joyful Health announces $17 million in Series A funding, bringing its total raised to $22 million.
Nym appoints Lori Jones (Aptarro) CEO.
AmplifyMD names Kurt Blasena (Digital Diagnostics) chief commercial officer.
Alameda Health System Director of Emergency Ultrasound Arun Nagdev, MD (Exo) joins Butterfly Network as chief medical officer.
UnitedHealthcare will eliminate most prior authorization requirements for 1,500 rural and critical access hospitals across all lines of business, and accelerate payments to those facilities by up to 50%.
Midland Health (TX) automates the patient assignment process for its hospitalist program using software from Medaptus.
VA Deputy Secretary Paul Lawrence, PhD says that the department’s go-lives in Michigan earlier this month went as expected, with no major issues reported thus far. He credits super users and pay-it-forward teams for ensuring a smooth transition to the new Oracle Health-based EHR. He summarizes, “This is the result of a lot of hard work. Sure, there are some hiccups, but so far everyone is feeling good. We haven’t let our foot off the gas. We’re still watching. We are still 24/7.”
Democratic lawmakers demand that the White House’s Office of Personnel Management drop its plan to collect detailed medical and pharmaceutical claims data on more than 8 million federal workers, retirees, and their families. Senate Democrats and the largest federal employee union argue that the proposal likely violates HIPAA, and that it raises concerns about health data potentially being used to target federal workers.
Kaiser Permanente researchers determine that overweight patients who actively engage in the provider’s app-based Lifestyle, Eating, and Activity in Pregnancy program have lower rates of weekly and total weight gain and are less likely to have babies with large weights for their gestational age. Study participants were equipped with digital scales and wearables, and had access to lifestyle coaches via text and phone.
Cerner co-founder Cliff Illig reminisces about the company’s early days, noting that friends didn’t think his partnership with the late Neal Patterson would last more than three weeks:
We ended up spending 40 years together, living out of the same pocketbook, working on everything. We were always focused on the bigger picture. Over the years, we had to work through a lot of stuff. I think Neal, literally, has said that he could count on one hand the number of times we disagreed to the point where one of us got up and walked out of the room. A lot of times our talks were over glasses of scotch, late in the evening, just sitting there with our feet up, watching the sunset, talking about what we were trying to accomplish. I encourage partnerships because it’s a lot easier when you’ve got somebody that you can share it with.
Blog Posts
Mr. H, Lorre, Jenn, Dr. Jayne.
Get HIStalk updates.
Send news or rumors.
Follow on X, Bluesky, and LinkedIn.
Sponsorship information.
Contact us.
ModMed Acquires Bonsai Health to Accelerate AI-Powered Patient Engagement
Ambulatory health IT vendor ModMed acquires patient engagement automation company Bonsai Health.
Coral, which offers automated document intake and data extraction solutions for radiology practices, infusion centers, specialty pharmacy, and DME suppliers, announces $12.5 million in funding.
Ethermed Raises $8.5 Million Series A
Philadelphia-based prior authorization automation startup Ethermed secures $8.5 million in Series A funding.
I’ve been in healthcare IT for a long time, so I am not shocked when I notice that I’m the oldest person in the room during a work meeting. When I was starting out as a physician informaticist, there was exactly one of me for a multi-hundred physician medical group. Our institution wasn’t terribly interested in paying physicians to be involved in technology projects, since leadership tied physician value solely to their ability to generate revenue.
It took more than a decade, and the insistence of a certain EHR vendor during an impending EHR conversion, for the organization to become willing to invest in clinical informatics.
Times have changed, and I am interacting with more members of Generation Z. As digital natives, they have always been connected with technology. This has created greater interest in clinical systems as they enter medical school, particularly for those who have worked as scribes or in clinical roles as part of the application process.
They have built upon these skills during their coursework. They have carried that interest into residency, often seeking opportunities to become EHR super-users, technology evaluators, or part of the technology build team.
Medical education during the last decade, at least at my alma mater, has recognized that learning styles vary. Different course materials are provided, including libraries of lecture videos that have eliminated the need for some students to attend class. Students are more focused on board scores, and many of them work with supplemental board preparation resources while consuming the lecture content at 2x speed.
I wanted to learn more about how this generation approaches learning. One of my favorite teachers recommended that I read “The Anxious Generation” by Jonathan Haidt.
The book summarizes the factors that influence differences across generations, including shared experiences. I immediately thought of the Vietnam War, the Cold War, and the Space Race, which I had heard about during childhood.
The book also notes that generations are defined by the technologies that they use in childhood, from radio, to television, and then from personal computers to the internet and smartphones.
Generation Z is the first to have their adolescence fully chronicled online, with the book noting that they are “the test subjects for a radical new way of growing up, far from the real-world interactions of small communities in which humans evolved. Call it the Great Rewiring of Childhood.”
The book also talks about the transition from a play-based childhood to a device-based childhood, including laptops, tablets, and smartphones. Communications have became disembodied and asynchronous, with increasing communication to a wide audience and within an online world where participants can just leave when they want without necessarily resolving conflicts.
It chronicles the rise of adolescent mental health issues, such as depression and anxiety, that parallel the rise in technology use. It covers the decline of what it calls “risky play” among children, and how that shifts young people from being in “discovery” mode to “defensive” mode. It discusses the idea of a psychological immune system and how humans need challenges to learn how to handle adverse events in a healthy way.
As a proud member of Generation X, one of my favorite parts of the book included pictures of playground equipment that would never be allowed today, each of which was present at my own elementary school.
The book talks about how smartphones and digital tools are often so interesting to users that they lose interest in non-screen-based activities. Anyone who has ever watched a friend or partner vegetate online, rather than participating in a conversation with the person in front of them, is familiar with this phenomenon.
It made me think about whether people who consume material online or through multimedia channels learn with the same level of depth as those who use traditional methods, such as reading or attending a lecture. I certainly wonder about that when I’m working with students who are using AI-based tools to look up clinical information, because anecdotally it feels like retention might be less than what I’m used to seeing with students I precept.
After reading about the decline in risk-taking behaviors among young people, I remembered one of the conversations that I recently had with some adjunct clinical faculty. The discussion was about bedside teaching rounds in the hospital, where traditionally a group of students, interns, and residents works with an attending physician to review cases of admitted patients. Questions are directed to students and trainees, which can be uncomfortable if you haven’t read up on the cases.
Debate followed about whether students should be allowed to consult phones during rounds, or if instead they should have to answer based solely on their knowledge and recall ability.
One colleague noted that students are more likely to answer questions if they are allowed to look up information, which they felt should be allowed because it parallels what senior physicians do when they don’t know the answer. There was back-and-forth about people who don’t look things up because they think that they know everything, which therefore makes them unsafe.
It was one of the more spirited discussions that I’ve been in this year. In hindsight, I wonder if students who don’t raise their hands aren’t lacking knowledge, but rather are avoiding the risk of being wrong.
The book also contains a deep discussion of the “foundational harms” that were caused by this Great Rewiring of Childhood, including social deprivation, sleep deprivation, attention fragmentation, and addiction. I’ve seen these among my pediatric patients. After reading the book, I am more likely to recognize signs of these in colleagues and coworkers, even among the older members of the team.
I’m curious how these elements could also be contributing to clinician burnout, especially since technology is deeply embedded in every aspect of patient care. Unless they’re in a strong call group where they can trust signing out to a colleague, and unless they also have the fortitude to avoid checking on patients when they’re not on call, it’s hard to get away from your phone.
The book closes with recommendations on what schools and governments can do to help counter the effects of increasing device use. These include limiting access to phones in schools, which is already happening in a number of states. The author also calls for governments and tech companies to address the issue by raising the age for teen internet use from 13 to 16 and for parents to be alert to signs of problematic use.
The book isn’t an easy read. The notes, footnotes, and index are over 80 pages. I would love to see a digest version that targets parents who are raising young children, although I’m not sure how well it would resonate given their own communal level of technology immersion.
What do you think about the idea of a Great Rewiring? Are you seeing examples of these phenomena in your workplace and in your families?. If you read the book, would you recommend it? What was your favorite takeaway? Do you recommend other books on the topic? Leave a comment or email me.
Email Dr. Jayne.
Optimizing Data Sharing in Rural Healthcare with a Data Lattice
By Jeff Brandes
Jeff Brandes is president and CEO of Azara Healthcare.
Focus is intensifying on securing the future of rural health. Across all 50 US states, the infusion of Rural Health Transformation Program (RHTP) dollars is intended to address rural health challenges through the deployment of workforce solutions, new access models, and sustainable technology infrastructure initiatives.
The question is whether the dollars allocated dollars will achieve the Centers for Medicare and Medicaid Services (CMS) goals of improving access and patient outcomes in a sustainable way without ongoing state and federal funding.
The stakes are high given the challenges of engaging complex, vulnerable populations in care. Success hinges on building an infrastructure that supports targeted interventions and outreach to prevent decline in patients with chronic conditions.
Shared data is critical. Just as physical infrastructures such as electricity and well-maintained roads are necessary for healthcare access, so is proactive collaboration between key stakeholders. Care teams across community-based hospitals, Federally Qualified Health Centers, behavioral health, public health, and other social service organizations need access to shared, up-to-date patient information to address health issues and close care gaps.
Optimal use of RHTP funds should start with the development of a shared “data lattice” among health centers, hospitals, and health plans to form the basis for healthcare infrastructure. State leaders can then devise a broader strategy that ensures that stakeholders are working from a single source of truth to achieve the goals of value-based care.
Digging Deeper: What is a Shared Data Lattice?
Whether it’s food deserts, lack of jobs, or limited education resources, rural communities often operate from a deficit. The same is true when it comes to patient data that is housed in today’s rural health organizations, many of which are part of the healthcare safety net caring for low-income, uninsured, and vulnerable populations.
Outdated legacy and EHR systems leave data in silos, which often necessitates manual documentation and referral processes that slow care delivery. The result is deferred care, poor and missed revenue opportunities or penalties from inaccurate reporting.
A unified data lattice among health centers, hospitals, and health plans offers a solution. It creates a normalized, shared layer of truth across organizations that have not traditionally functioned as a coordinated network.
In this scenario, primary care can share a patient’s accurate medication history with a behavioral health provider down the road, for example. Or coordinated views between health plans, primary care, and social service agencies can help identify patients who need colorectal cancer screening but do not have transportation, which enables targeted interventions to help them access care.
Building this connected rural healthcare ecosystem requires data integration that supports aggregation of clinical data from disparate EHRs, Medicaid and Medicare Advantage claims, referral and admission and discharge feeds, social drivers of health (SDOH), and care management insights within care team workflows. In addition, systems must support automated regulatory and value-based reporting.
The key to success is alignment between partners who share patients and a commitment to innovation. In rural communities, where roughly 20% of Americans live, investments in shared data infrastructure will reduce the burden on clinicians at a time when workforce is at a premium. Meaningful investments in generative AI, not the next shiny object, can accelerate data transformation and reporting tasks.
Shared Data Lattice: Making the Business Case
State leaders will soon need to demonstrate ROI and sustainability for their RHTP investment dollars. That accountability requires more than good intentions. It must make the business case for the right technology infrastructure to power better care delivery and better reporting.
Without a unified data infrastructure, calculating ROI is often slow and often requires manual processes to pull together needed sources and to ensure that reporting is reliable. In contrast, a shared data lattice establishes a framework for clarity.
For example, when stakeholders can tie avoided emergency department visits to dollars invested in preventive care in near real-time, that’s hard ROI. Or, when risk stratification processes lead to better identification of diabetes patients at risk of decline, state leaders can link interventions to improved A1C scores.
For public health and advocacy organizations, the value of a shared data lattice extends beyond operations. It replaces siloed reporting with real-time intelligence that supports both prevention and coordinated care. States can monitor progress continuously and adjust strategy proactively instead of reacting to crises. With advanced infrastructure, legislators can track progress through rural impact scorecards that aggregate metrics across health centers, hospitals, and health plans.
For rural health centers and hospitals, the benefits are immediate, as they eliminate data blind spots. Organizations can demonstrate value using hard evidence, and alignment with state rural transformation becomes more likely.
Data: An Essential Community Health Infrastructure
A rural health data lattice ensures that independent providers have the same visibility, coordination, and analytics capabilities that large health systems routinely rely upon. It ensures that the most vulnerable patient populations have the same quality access to care as others.
Measurable rural transformation does not have to be distant. With the right infrastructure and partnerships, it can begin now.
Sectra completes Oxipit acquisition, accelerating autonomous AI in radiology
Medical imaging vendor Sectra acquires Oxipit, which offers AI solutions such as autonomous chest x-ray analysis.
Siemens to hold Healthineers spin-off vote in February 2027
Siemens will vote on whether or not to spin off its Healthineers business in February 2027.
Ōura Welcomes Galen AI Team, Acquires Technology to Advance Connected Health AI Roadmap
Smart ring company Oura acquires AI-based personal health companion startup Galen AI.
Medical imaging vendor Sectra acquires Oxipit, which offers AI solutions such as autonomous chest x-ray analysis.
Respondents to last week’s poll believe that ambient scribe tools will give clinicians a headline benefit, but also allow vendors to cash in while health systems quietly ponder converting efficiency into more billables. Hint: history suggests that any healthcare technology innovation will quickly become a revenue factory, because people and organizations consistently take whatever action delivers the greatest benefit to themselves.
New poll to your right or here: If AI reduces patient visits, what changes most?
None scheduled soon. Contact Lorre to have your resource listed.
UCI Health promotes Deepti Pandita, MBBS to chief medical and informatics officer.
PerfectServe names Steve Tyler (TigerConnect) as CTO.
Fulton County Medical Center (PA) will join WVU Health, with the hospital saying that a key part of the agreement is that it will move from Meditech to WVU’s Epic system.
In Australia, Justice Health NSW goes live on the Epic-powered Single Digital Patient Record, becoming the first NSW Health organization to do so. The system will eventually link 220 hospitals, 150 pathology collection centers, 65 labs, and 600 community health centers across New South Wales.
Priority Dispatch launches a SMART on FHIR app that connects the company’s emergency dispatch protocols to Epic so that hospitals can coordinate inter-facility patient transfers. The Medical Transfer Protocol app was built using VectorCare’s Smart on FHIR as a Service platform.
A New York physician pleads guilty to orchestrating a scheme that fraudulently billed insurers tens of millions of dollars for COVID-19 testing and related services, including submitting false claims and fabricated medical records, causing $24 million in losses. Anesthesiologist Ali Rashan, MD opened several ClearMD clinics in New York City during the pandemic, staffing them with untrained college-aged “medical assistants” who fed samples into a machine, which then generated patient emails that promised a physician examination and telehealth visit that never happened. Prosecutors allege that he instructed employees to create a program to fabricate progress notes and test results to support the fraudulent billing.
Stripe CEO Patrick Collison says that the most useful preventive medical advice that he has received came from using AI coding agents to analyze his genome to recommend specific tests and treatments. He says that for a few hundred dollars to sequence his genome and perform the analysis, he found that he was 30x predisposed to melanoma that may be prevented by supplements.
I asked Claude to list the exact steps that would be needed to replicate the experience above and to show me a sample report. I would be interested in having a clinician try this and let me know the result. Also, I’m wondering how many primary care doctors would be comfortable creating an action plan from a report like this one. I can see someone using Claude to create a business plan, e-commerce website, and marketing campaign in selling the entire genomics package for maybe $3,000 to become one of those “one person and AI” businesses that generate a fortune.
Robert Wachter, MD analyzes Epic’s dominance in healthcare AI, which he attributes to not necessarily offering the best tools, but benefiting from EHR integration, health system trust, and perceived long-term stability as health systems try to reduce IT risk and complexity. He also concludes that “good enough” beats “best” when choosing between Epic and a startup, that CIOs make job-safe choices, and that Epic has a strong distribution advantage in being able to quick-ship AI directly into clinician workflows.
Parkview Health SVP/CMIO Mark Mabus, MD, RPh advises health systems to say more than just “we use Epic” in provider recruitment and instead explain why their implementation is different.
The first episode of New York Times Opinion’s “Divided” program, “Who Is the US Health System Really For?” features a plastic surgeon and the former chief medical officer of CVS Health debate the question of whether insurance companies prioritize profits over patient care.
Blog Posts
Mr. H, Lorre, Jenn, Dr. Jayne.
Get HIStalk updates.
Send news or rumors.
Follow on X, Bluesky, and LinkedIn.
Sponsorship information.
Contact us.
Digital Behavioral Health Coordination Platform Trayt Health Lands $7M
Behavioral health and primary care coordination company Trayt Health announces $7.17 million in new funding.
Services at Brockton hospital return to normal more than a week after cyberattack
Signature Healthcare (MA) brings its digital systems back online and resumes normal operations after experiencing a ransomware earlier this month.
Joyful Health Raises $22M to Help Healthcare Providers Recover Billions in Unpaid Insurance Claims
Revenue intelligence software startup Joyful Health announces $17 million in Series A funding, bringing its total raised to $22 million.
The Department of Justice accuses telehealth startup Zealthy of widespread fraud, including issuing prescriptions using doctors’ identities without consent, using deceptive billing practices, and misusing customer data.
The DOJ seeks an asset freeze and receivership, adding to its original 2024 complaint. Regulators warn that financial penalties and required consumer refunds could exceed the company’s resources and may force it into bankruptcy.
Zealthy CEO Kyle Robertson was fired from Cerebral in 2022 after the company was fined for unauthorized distribution of Adderall, and prosecutors allege that “Robertson’s lawbreaking is only becoming more brazen and dangerous.”
From Front Door Fred: “Re: digital front doors. Hospitals spent millions on them, and now vendors are inserting an AI receptionist in front of them to direct the patient. Why turn that over to a third party?”
From Oiler: “Re: ChatGPT health advice. If patients are skipping visits because ChatGPT told them they are fine, are health systems losing customers or just low-margin visits they didn’t really want? The payer still decides who gets paid.”
Ms. M says that her Clearlake, CA kindergarten class is using the math books that HIStalk readers provided in funding her Donors Choose grant request, with matching funds from my Anonymous Vendor Executive.
None scheduled soon. Contact Lorre to have your resource listed.
UnityPoint Health will eliminate 207 IT and RCM positions and outsource their work to third-party vendors.
Abridge adds NEJM and the JAMA Network to its point-of-care clinical evidence.
An analysis of de-identified Microsoft Copilot conversations finds that health AI is being widely used for personal symptom assessment, condition questions, and emotional support, with usage peaking at night and often substituting for unavailable care. One in seven inquiries are asked on behalf of someone else, which suggests that it is being used by caregivers or relatives.
WellSky adds AI capabilities to its CarePort provider directory for post-acute care discharge decisions.
Hippocratic AI releases voice AI products for call centers and inpatient nurses.
Care coordination platform vendor One Call acquires Data Dimensions, which offers EDI and clearinghouse solutions.
Allbirds, which recently sold its formerly trendy wool shoe business to a brand name acquisition company for a paltry $39 million versus the company’s one-time market value of $4 billion, renames itself as NewBird AI as it pivots to renting AI infrastructure. Shares jumped 600% on the news, leading one market strategist to conclude, “A 6x or 7x move for a company that is literally ditching its prior business model for one in which it has no demonstrated expertise says quite a bit about a market froth and investor willingness to chase moves.”
Blog Posts
Mr. H, Lorre, Jenn, Dr. Jayne.
Get HIStalk updates.
Send news or rumors.
Follow on X, Bluesky, and LinkedIn.
Sponsorship information.
Contact us.
People often ask me what I think of the different AI tools. I regularly use a host of them via both free and paid accounts.
I received an email the other day from Doximity that told me that the platform had “compiled year-over-year prescription data” in my area “to help discover meaningful trends in treatment.” Clicking the button took me to a preloaded search within DoxGPT that was based on the prompt, “Create a report highlighting trends in prescription treatment in my area versus nationally.” The preamble to the response noted that its goal is to highlight shifts in local practice patterns relative to national movement.
I have spent my professional career looking at data and assessing the validity of different kinds of evidence. I struggle to understand how this kind of information would be useful to me. It felt like AI for AI’s sake.
I don’t order labs or tests for patients unless the results could change how I manage them. I didn’t ask for this AI query and it’s not obvious what I should be doing with it, so why bother asking? Knowing which brands of GLP-1 drugs I’m prescribing relative to my peers isn’t helpful. Those decisions are usually driven by the patient’s insurance coverage.
The statistics that were presented didn’t give any indication as to their quality. One drug was down 83% locally versus 88% nationally, but without more information, I have no way of knowing whether this is a statistically significant change. Another drug was down 17% locally versus 16% nationally, which is meaningless.
A company thinking that I want this information tells me that they don’t understand their audience and don’t know what is valuable to a physician in my specialty. They monetized my click when I went to the site to view the content, and I knew that before I decided to take a look, but it’s just baffling that they serve this information to physicians who are strapped for time and cognitive bandwidth.
The CEO of NYC Health and Hospitals is under fire for comments that he made at a business conference. President and CEO Mitchell Katz stated, “We could replace a great deal of radiologists with AI at this moment, if we are ready to do the regulatory challenge.” He used the example of breast cancer screening, saying that radiologists should only weigh in when AI systems identify abnormal images.
As a physician, my first thought was that this person probably doesn’t understand concepts such as false-positives and false-negatives. The article also mentions the concept of an AI mirage, which goes beyond hallucinations and can bypass hallucination safeguards by providing rational explanations.
Another panelist noted that his hospital is already using similar technology. He clarified that the system is being used for women who aren’t high risk, and even then, the false-negative rate is three out of 10,000. Published reports of his comments don’t mention what the false-negative rate is for human radiologists at the center.
A radiologist who is critical of the proposal, Mohammed Suhail, MD, said that the statements are “undeniable proof that confidently uninformed hospital administrators are a danger to patients… Hospitals are happy to cut costs even if it means patient harm, as long as it’s legal.”
The so-called mirage reasoning effect is discussed in a preprint journal article. Stanford University researchers describe AI models that created “detailed image descriptions and elaborate reasoning traces, including pathology-biased clinical findings, for images never provided.”
The authors believe that the AI models use their memory and language skills to hide their weaknesses. They note that they saw this behavior in models from OpenAI, Google, and Anthropic. It’s yet another example of how AI can be very convincingly wrong and why we need to remember that tools are not without risk.
The American Medical Informatics Association has joined the National Health Council as a Partner of Patient Organizations Member. The Council, which was founded in 1920, provides advocacy for the 200 million people who are living with chronic diseases and disabilities, along with their family caregivers. AMIA will share its research, policy, and clinical expertise to promote shared goals, including priorities in health equity, access, innovation, and artificial intelligence. AMIA will join 170 national organizations that support the Council.
As the conflict with Iran continues, one thing I didn’t have on my bingo card for the year is a disruption in the global helium supply that creates risk for healthcare organizations. Helium is used to cool the magnets that are found in MRI scanners and is also used for some laparoscopic and subspecialty surgical procedures. Production facilities in Qatar have been damaged and shipping has been disrupted.
The helium supply has been unstable for some time. Imaging vendors are developing low-helium scanners, but those aren’t widely used. I’m curious to hear from provider-side readers whether their organizations are discussing the issue and whether they are seeing gaps in the supply chain.
From Utterly Presumptuous: “Re: conferences that sell their attendee lists with no opt-out for marketing spam. I work for a healthcare-adjacent company and attend many of the health IT conferences. I received more than 20 emails requesting meetings at a recent conference. Some of them make assumptions that because I’m attending a health IT conference that I’m part of a provider organization. The most annoying one I received today started with, ‘Since your company is already using our services, X person at our organization wanted to meet you at the show.’ I own all the contracts, so I know for a fact that we are not doing business with them. Based on their tactics, I don’t see myself doing business with them in the future.”
I have worked with vendors who subscribe to the philosophy that being told no is still a response, but that doesn’t make these direct emails less annoying. I make liberal use of the “mark as junk” button in my email client, but I’m not sure that it reduces the volume. If others have tips, let me know.
Does it annoy you to receive emails that were created from conference attendee lists, or does your company see benefit from sending them? Leave a comment or email me.
Email Dr. Jayne.
HIStalk Sponsor Announcements Industry Events Webinars
TBF the anti-hate non profit was funding undercover investigators trying to being down the KKK & has done successfully for…