THCB

Categories

Category: Health Tech

China Goes “Democratic” on Artificial General Intelligence

By MIKE MAGEE

Last week, following a visit to the White House, Jensen Huang instigated a wholesale reversal of policy from Trump who was blocking Nvidia sales of its H20 chip to China. What did Jensen say?

We can only guess of course. But he likely shared the results of a proprietary report from noted AI researchers at Digital Science that suggested an immediate policy course correction was critical. Beyond the fact that over 50% of all AI researchers are currently based in China, their study documented that “In 2000, China-based scholars produced just 671 AI papers, but in 2024 their 23,695 AI-related publications topped the combined output of the United States (6378), the United Kingdom (2747), and the European Union (10,055).”

David Hook, CEO of Digital Science was declarative in the opening of the report, stating “U.S. influence in AI research is declining, with China now dominating.”

China now supports about 30,000 AI researchers compared to only 10,000 in the US. And that number is shrinking thanks to US tariff and visa shenanigans, and overt attacks by the administration on our premier academic institutions.

Economics professors David Autor (MIT) and Gordon Hanson (Harvard), known for “their research into how globalization, and especially the rise of China, reshaped the American labor market,” famously described the elements of “China Shock 1.0.” in 2013. It was “a singular process—China’s late-1970s transition from Maoist central planning to a market economy, which rapidly moved the country’s labor and capital from collective rural farms to capitalist urban factories.”

As a result, a quarter of all US manufacturing jobs disappeared between 1999 and 2007. Today China’s manufacturing work force tops 100 million, dwarfing the US manufacturing job count of 13 million. Those numbers peaked a decade ago when China’s supply of low cost labor peaked. But these days China is clearly looking forward while this administration and its advisers are being left behind in the rear view mirror.

Welcome to “China Shock 2.0” wrote Autor and Hanson in a recent New York Times editorial. But this time, their leaders are focusing on “key technologies of the 21st century…(and it) will last for as long as China has the resources, patience and discipline to compete fiercely.”

The highly respected Australian Strategic Policy Institute, funded by their Defense Department, has been tracking the volume of published innovative technology research in the US and China for over a quarter century. They see this as a measure of experts opinion where the greatest innovations are originating. In 2007, we led China in the prior four years in 60 of 64 “frontier technologies.”

Two decades later, the table has flipped, with China well ahead of the US in 57 of 64 categories measured.

Continue reading…

Healthcare AI: What’s in your chatbot?

By OWEN TRIPP

So much of the early energy around generative AI in healthcare has been geared toward speed and efficiency: freeing doctors from admin tasks, automating patient intake, streamlining paperwork-heavy pain points. This is all necessary and helpful, but much of it boils down to established players optimizing the existing system to suit their own needs. As consumers flock to AI for healthcare, their questions and needs highlight the limits of off-the-shelf bots — and the pent-up demand for no judgment, all-in-one, personalized help.

Transforming healthcare so that it actually works for patients and consumers — ahem, people — requires more than incumbent-led efficiency. Generative AI will be game-changing, no doubt, but only when it’s embedded and embraced as a trusted guide that steers people toward high-quality care and empowers them to make better decisions.

Upgrading Dr. Google

From my vantage point, virtual agents and assistants are the most important frontier in healthcare AI right now — and in people-centered healthcare, period. Tens of millions of people (especially younger generations) are already leaning into AI for help with health and wellness, testing the waters of off-the-shelf apps and tools like ChatGPT.

You see, people realize that AI isn’t just for polishing emails and vacation itineraries. One-fifth of adults consult AI chatbots with health questions at least once a month (and given AI’s unprecedented adoption curve, we can assume that number is rising by the day). For most, AI serves as a souped-up, user-friendly alternative to search engines. It offers people a more engaging way to research symptoms, explore potential treatments, and determine if they actually need to see a doctor or head to urgent care.

But people are going a lot deeper with chatbots than they ever did with Dr. Google or WebMD. Beyond the usual self-triage, the numbers tell us that up to 40% of ChatGPT users have consulted AI after a doctor’s appointment. They were looking to verify and validate what they’d heard. Even more surprising, after conferring with ChatGPT, a similar percentage then re-engaged with their doctor — to request referrals or tests, changes to medications, or schedule a follow-up.

These trends highlight AI’s enormous potential as an engagement tool, and they also suggest that people are defaulting to AI because the healthcare system is (still) too difficult and frustrating to navigate. Why are people asking ChatGPT how to manage symptoms? Because accessing primary and preventive care is a challenge. Why are they second-guessing advice and prescriptions? Sadly, they don’t fully trust their doctor, are embarrassed to speak up, or don’t have enough time to talk through their questions and concerns during appointments.

Chatbots have all the time in the world, and they’re responsive, supportive, knowledgeable, and nonjudgmental. This is the essence of the healthcare experience people want, need, and deserve, but that experience can’t be built with chatbots alone. AI has a critical role to play, to be sure, but to fulfill its potential it has to evolve well beyond off-the-shelf chatbot competence.

Chatbots 2.0

When it comes to their healthcare, the people currently flocking to mass-market apps like ChatGPT will inevitably realize diminishing returns. Though the current experience feels personal, the advice and information is ultimately very generic, built on the same foundation of publicly available data, medical journals, websites, and countless other sources. Even the purpose-built healthcare chatbots in the market today are overwhelmingly relying on public data and outsourced AI models.

Generic responses and transactional experiences have inherent shortcomings. As we’ve seen with other health-tech advances, including 1.0 telehealth and navigation platforms, impersonal, one-off services driven primarily by in-the-moment-need, efficiency, or convenience don’t equate to long-term value.

For chatbots to avoid the 1.0 trap, they need to do more than put the world’s medical knowledge at our fingertips.

Continue reading…

Dominique Wells, Conduit Health Partners

Dominique Wells is COO of Conduit Health Partners which is a spin off from the (now) Bon Secours Mercy Health system. Their role is to provide back up for nursing staff for health systems in very specific areas, notably patient transfer operations, nurse triage and patient communications. Dominique and her team showed me a brief demo of how the transfer operation works. We also got into the conversation about the role of AI in nursing, and how nursing has changed since the pandemic. An interesting discussion about how the most vital role in health care is changing and how new services are being developed to adapt to it—Matthew Holt

Watching Where and How You’re Walking

By MIKE MAGEE

In a speech to the American Philosophical Society in January, 1946, J. Robert Oppenheimer said, “We have made a thing …that has altered abruptly and profoundly the nature of the world…We have raised again the question of whether science is good for man, of whether it is good to learn about the world, to try to understand it, to try to control it, to help give to the world of men increased insight, increased power.”

Eight decades later, those words reverberate, and we once again are at a seminal crossroads. This past week, Jensen Huang, the CEO of Nvidia, was everywhere, a remarkably skilled communicator celebrating the fact that his company was now the first publicly traded company to exceed a $4 trillion valuation.

As he explained, “We’ve essentially created a new industry for the first time in three hundred years. the last time there was an industry like this, it was a power generation industry…Now we have a new industry that generates intelligence…you can use it to discover new drugs, to accelerate diagnosis of disease…everybody’s jobs will be different going forward.”

Jensen, as I observed him perform on that morning show, seemed just a bit overwhelmed, awed, and perhaps even slightly frightened by the pace of recent change. “We reinvented computing for the first time since the 60’s, since IBM introduced the modern computer architecture… its able to accelerate applications from computer graphics to physics simulations for science to digital biology to artificial intelligence. . . . in the last year, the technology has advanced incredibly fast. . . AI is now able to reason, it’s able to think… Before it was able to understand, it was able to generate content, but now it can reason, it can do research, it can learn about the latest information before it answers a question.”

Of course, this is hardly the first time technology has triggered flashing ethical warning lights. I recently summarized the case of Facial Recognition Technology (FRT). The US has the largest number of closed circuit cameras at 15.28 per capita, in the world. On average, every American is caught on a closed circuit camera 238 times a week, but experts say that’s nothing compared to where our “surveillance” society will be in a few years.

The field of FRT is on fire. 

Continue reading…

Towards a Tricoder

By MIKE MAGEE

On March 9, 1967, the Star Trek classic episode, “The Devil in the Dark” first aired. The Enterprise had received an urgent distress call from miners on the planet Janus VI. They are literally melting after, Horta, a wounded inhabitant has targeted them with liquifying acid rays.

A sympathetic Spock hears the call, and in an effort to disclose cause and motivation, “mind-melts” with the creature. Turns out, all she’s trying to do is protect her babies from a perceived threat. Kirk agrees, and with Spock, calls in Dr. McCoy to access the patient’s condition.

What McCoy encounters is a “rocky-skinned patient.” With the aid of his tricoder, a handheld diagnostic sensor, “Bones” (McCoy’s nickname referencing the historical 19th century American slang “Sawbones” referring to surgeons) uncovers a serious and deep gaping wound that requires immediate attention.

Kirk manages to “beam down” a hundred pounds of thermoconcrete, and McCoy expertly applies it to the wound. All of which is a set-up for his shipmates to wonder if this will work, which generates the iconic most-repeated line in the series storied history. McCoy (clearly irritated) utters – “How do I know? I’m a doctor, not a bricklayer.”

Similarly challenged modern day doctors have been voicing their own frustrations for more than a few decades. But the AMA has been scientifically tracking their discontent only since 2011. The levels of burnout are somewhat down in 2025 compared to peaked pique in 2021. But among the irritants, integration of new technology remain near the top of the list.

Continue reading…

Matthew tries Reperio’s at home health screening

We are entering an age of at home testing and the team at Reperio just raised $14m to make weight, blood pressure and cholesterol/blood sugar testing available at home. But this is a relatively complex series of tests, intended to get people who haven’t been to a primary care doctor back into the system. How is the experience and can we expect people to do it? And does the result correlate with standard lab testing? They sent me the box for me to find out. I totally screwed it up the first time (apparently only 4% of people do), but they gave me another chance. So come along with me to find out how it works. Would you do this, or just go to Labcorp?! — Matthew Holt

BTW since I did this Reperio released an early customer study that said 23% of those who used the kit found a problem they weren’t aware of. Which is I guess the point!

Henry Ford, the Model T, and Digital Health

By TREVOR VAN MIERLO

Most of us know the story of the Model T – but what’s often overlooked is how it applies to other industries, especially digital health. Let’s revisit:

In the early 1900s cars were custom built. You’d meet with a consultant, design the car, place your order, and wait for months for delivery. Once your car was delivered, it was difficult to operate. Many owners hired chauffeurs because at the time, cars required technical know-how, constant maintenance, and a fair bit of strength (power steering wouldn’t arrive for decades).

Then came the Model T in 1908, which led to Ford developing his assembly line in 1913. He recognized a problem in the industry and saw an opportunity. He saw the opportunity for scale:

  • Standardization: “Any color, as long as it’s black“
  • Mass production = affordability: Prices dropped from $850 in 1908 to $300 within a decade
  • Accessible ownership: anyone could walk into a Ford dealership and drive away 
Left: Early car assembly (pre-Ford) Right: A leading digital health interface, 2000

On the right side of the above image is a cutting-edge digital health program from August 2000. I know it well – because I helped build it. Since then, I’ve worked on well over 100 digital health interventions. Probably closer to 200. Here’s the thing: what’s inside hasn’t changed very much. Behavioral science doesn’t move that quickly (although my recent work in AI is changing that).

And yes – digital interventions look better, are easier to navigate, and coding languages have evolved – but practically, digital health is still building custom cars – not Model Ts. That’s why tens of millions can’t open a browser and get the help they need.

What’s Blocking Digital Health’s Model T Moment?

1. Enterprise Sales (Death by Pipeline): Most digital health tools are sold through enterprise channels: RFPs, procurement departments, tenders, security reviews, and legal teams. The average sales cycle is 6-18 months. That’s fine for a $5M contract, but it’s lethal for a $50,000 contract. The problem isn’t the product – it’s the process.

2. The Vanishing Champion: I’ve experienced this dozens of times, and I’ve taken deep breaths watching it unfold on webinars: a digital health company demos their solution alongside a client champion. Priorities shift. The champion leaves. The reference project dies. Most contracts aren’t lost on merit – they’re lost to turnover.

3. Healthcare Pricing ≠ Software Pricing: Most patient-facing tools are priced like services, not products. That’s a symptom of the enterprise sales trap. Vendors charge annual fees regardless of usage. Clients expect hand-holding for these custom products. Pricing needs to reflect modern SaaS models – freemium, tiered access, per-user billing.

4. Static Products in a Dynamic World: Consumer software updates weekly – sometimes daily. Digital health tools? They launch, then stall. Feedback loops are weak. There’s no culture of iteration, and no expectation of continuous improvement.

5. Nobody Markets to the User: The best-designed tools fail if no one uses them. Lack of engagement is a systemic issue, yet many programs are launched without onboarding plans, email campaigns, or even prewritten content for TikTok or Instagram. Users don’t know what the tool is, why they received access to it, how they access it, or how it fits into their care. That’s not a product issue – it’s a marketing failure.

We Need to Build the Systems, Not Just the Tool

Henry Ford didn’t invent the automobile, but he’s remembered because he built a system. He looked beyond the engine, the chassis, and the tires. He focused on standardization, distribution, and access.

Digital health needs the same. Right now, too many solutions are trapped in a loop – custom-built for small populations, sold through enterprise channels, with no realistic path to scale.

The Good News? We’re Close

Cloud infrastructure, AI, and behaviorally intelligent platforms are finally catching up. We can now personalize at scale, launch instantly, track engagement in real time, and iterate fast. But to get there, we have to let go of the custom-built carriage mindset and embrace the assembly line. That’s not a compromise in quality – it’s a commitment to reach.

  • We don’t need more pilots – we need platforms.
  • We don’t need more bespoke builds – we need scale.

Digital health doesn’t have a technology problem – it has a delivery problem.

Until we achieve that, we’re just making nicer carriages – while the world waits for its Model T.

Dr. Trevor van Mierlo has built mental health and patient support products for more than two decades and is the CEO of Evolution Health

A New Future for DNA

By KIM BELLARD

As a DNA-based creature myself, I’m always fascinated by DNA’s remarkable capabilities. Not just all the ways that life has found to use it, but our ability to find new ways to take advantage of them. I’ve written about DNA as a storage medium, as a neural network, as a computer, in a robot, even mirror DNA. So when I read about the Synthetic Human Genome (SynHG) project, last month, I was thrilled.   

The project was announced, and is being funded, by the Wellcome Trust, to the tune of £10 million pounds over five years. Its goal is “to develop the foundational tools, technology and methods to enable researchers to one day synthesise genomes.”

The project’s website elaborates:

Through programmable synthesis of genetic material we will unlock a deeper understanding of life, leading to profound impacts on biotechnology, potentially accelerating the development of safe, targeted, cell-based therapies, and opening entire new fields of research in human health. Achieving reliable genome design and synthesis – i.e. engineering cells to have specific functions – will be a major milestone in modern biology.

The goal of the current project isn’t to build a full synthetic genome, which they believe may take decades, but “to provide proof of concept for large genome synthesis by creating a fully synthetic human chromosome.”

That’s a bigger deal than you might realize.

“Our DNA determines who we are and how our bodies work,” says Michael Dunn, Director of Discovery Research at Wellcome. “With recent technological advances, the SynHG project is at the forefront of one of the most exciting areas of scientific research.” 

The project is led by Professor Jason Chin from the Generative Biology Institute at Ellison Institute of Technology and the University of Oxford, who says: “The ability to synthesize large genomes, including genomes for human cells, may transform our understanding of genome biology and profoundly alter the horizons of biotechnology and medicine.”

He further told The Guardian: “The information gained from synthesising human genomes may be directly useful in generating treatments for almost any disease.”

Professor Patrick Yizhi Cai, Chair of Synthetic Genomics at the University of Manchester boasted: “We are leveraging cutting-edge generative AI and advanced robotic assembly technologies to revolutionize synthetic mammalian chromosome engineering. Our innovative approach aims to develop transformative solutions for the pressing societal challenges of our time, creating a more sustainable and healthier future for all.”

Project member Dr Julian Sale, of the MRC Laboratory of Molecular Biology in Cambridge, told BBC News the research was the next giant leap in biology: “The sky is the limit. We are looking at therapies that will improve people’s lives as they age, that will lead to healthier aging with less disease as they get older. We are looking to use this approach to generate disease-resistant cells we can use to repopulate damaged organs, for example in the liver and the heart, even the immune system.”

Consider me impressed.

Continue reading…

Anmol Madan, RadiantGraph

Anmol Madan is CEO of RadiantGraph. He’s building an end to end solution that goes from data ingestion to applications to consumer connection via text/email and voice in order to let payers quickly roll out patient engagement plans. His idea is that plans/payers don’t need to fix their data, RadiantGraph’s AI can take the messy data and and then add an AI layer, and on that create specific applications–Anmol showed me a comprehensive demo. I also asked him if they are doing too much, or conversely if they need to do more!–Matthew Holt

assetto corsa mods