Best of LinkedIn: Health Tech CW 44/ 45

00:00:00: Welcome to the deep dive.

00:00:01: We're digging into the data this week, really focusing on the key shifts happening in health tech.

00:00:05: We looked at a calendar when it's forty four and forty five.

00:00:09: And what we're seeing is, well, it's a fundamental change in how the industry measures success.

00:00:13: The common thread seems to be innovation moving past that, you know, easy pilot stage.

00:00:18: This deep dive is provided by Thomas Algeyer and Frennis, based on the most relevant LinkedIn post on health tech in CW forty four and forty five.

00:00:26: Frennis equips product and strategy teams with market and competitive intelligence to navigate this evolving landscape.

00:00:33: you know, cool technology and isolation.

00:00:38: We're really focused on applied innovation, how things like AI, connected monitoring, interoperability, how they're actually being integrated into practical workflows.

00:00:46: And crucially, the market is demanding, you know, it's demanding a clear demonstrable shift toward durable economics and actual health outcomes.

00:00:53: We're moving far beyond those vanity metrics.

00:00:56: Yeah, that focus on durable economics feels like a massive correction.

00:01:00: And it really sets up our first major theme.

00:01:02: Shifting the scoreboard, right?

00:01:04: Moving away from just chasing growth metrics like user numbers or big valuations towards measuring real clinical outcomes, right?

00:01:12: And that perspective is being well enforced by the market now.

00:01:16: The analyst Ganesh Kamath observed this pivot clearly.

00:01:19: He noted the global digital health equity funding actually fell by about twenty one percent quarter on quarter in Q to twenty twenty

00:01:27: five.

00:01:27: It's a bit of a sobering reality check, isn't it?

00:01:29: Even telehealth, you know, which peaked massively during the pandemic seems to have stabilized.

00:01:33: It's holding steady at around thirteen to seventeen percent of all visits now.

00:01:37: Sustainable, maybe?

00:01:38: And you can really feel the pain points when you look at certain case studies.

00:01:41: It takes sero-medical in Australia.

00:01:44: They had this really sophisticated epilepsy monitoring tech, strong science behind it, yet they ultimately shut down the hard lesson there.

00:01:51: It seems hospitals just weren't ready to pay for it, and insurance coverage simply wasn't lined up.

00:01:55: Exactly.

00:01:56: It highlights that strong science is necessary, absolutely, but it's just not sufficient if you don't have a sustainable business model baked in.

00:02:04: It's the difference between just having novel technology versus reliable health care delivery, isn't it?

00:02:09: Precisely.

00:02:10: And you can contrast that story with, say, Prachto in India.

00:02:14: Their success came from focusing on improving the actual healthcare process.

00:02:18: Things like managing bookings, reports, follow-ups, you know, the fundamentals.

00:02:23: That focus on foundational delivery that translated into steady profitable growth for them across more than twenty countries.

00:02:30: It's a different approach.

00:02:31: And this requires a pretty big strategic mindset shift for leaders, too.

00:02:35: I saw Medtronic.

00:02:35: CMO Austin Chang had some advice on this.

00:02:38: He called financial literacy a leadership multiplier.

00:02:41: For clinical leaders, this means being able to translate patient quality improvements directly into, you know, P&L and operating margin language.

00:02:49: That makes sense.

00:02:50: Growth isn't just about ticking every technical feature box anymore.

00:02:53: It's about making strategic bets on capacity and talent that actually generate profit and deliver value.

00:02:58: And

00:02:59: that economic pressure, it isn't limited to digital services either, right?

00:03:02: Not

00:03:02: at all.

00:03:04: You're finally highlighted a major slowdown, though possibly temporary, in the surgical robotics sector.

00:03:09: After years of really high growth, hospitals are starting to balk at buying new systems.

00:03:14: You've got immense capital costs, lower procedural volumes than expected, and uncertain reimbursement pathways.

00:03:22: So it sounds like the established players might have made the hardware just too expensive, too complex.

00:03:27: That's the implication.

00:03:29: Your final lease suggests this slowdown actually creates an opening.

00:03:32: It's the perfect time for new entrants to challenge the status quo.

00:03:36: They can come in with more agility, maybe better affordability, proposing things like modular pricing, leasing, or even paper use options.

00:03:43: Hmm,

00:03:44: paper use.

00:03:45: I wonder how feasible that really is for, you know, those massive risk averse US health systems.

00:03:49: They budget annually, right?

00:03:50: Doesn't that just shift the risk onto the vendor?

00:03:52: Well,

00:03:53: yes, it absolutely shifts the risk.

00:03:55: But the source material suggests this might be precisely what the market is demanding right now.

00:04:00: just to ease that immediate capital pressure.

00:04:02: When a hospital is staring down a huge capital outlay, easing that initial shock with flexible models, maybe combined with strong integration support, that could be as valuable as the innovation itself.

00:04:13: Okay, so if the financials are demanding hardware gets cheaper and smarter, well, the same intensity seems to be hitting the software world.

00:04:20: which brings us neatly to our second theme, the AI workflow revolution, particularly in delivery and diagnostics.

00:04:27: Right.

00:04:28: We've definitely moved past the proof-of-concept phase for AI.

00:04:31: The conversation now is almost entirely focused on integration.

00:04:35: How do we embed this intelligence directly into the clinician's workflow?

00:04:38: Make it boost precision, speed, efficiency.

00:04:41: And nothing really screams push for scale more than that big partnership between Verily and NVIDIA, does it?

00:04:47: According to Myung Cha and Nick Melrose, integrating NVIDIA's AI stack into Verily's pre-precision health platform, it's all about accelerating AI at scale, making it production-ready essentially.

00:04:58: Yeah, and they're even pushing into the consumer space now with that app, VerilyMe, featuring the AI companion violet.

00:05:05: Interesting move.

00:05:06: It

00:05:06: really shows that transition from the lab to, well... operational reality.

00:05:11: That's where the real impact

00:05:12: happens.

00:05:13: Exactly.

00:05:13: And look at interventional care.

00:05:14: Phillips launched a device guide.

00:05:16: It's an AI-powered co-pilot for cardiologists.

00:05:20: Atul Gupta explained how it uses real-time image fusion.

00:05:23: It actually guides physicians during complex, minimally invasive procedures, like mitral valve repair.

00:05:30: Wow.

00:05:30: Yeah, he used this great analogy describing the challenge of Navigating a catheter is like hitting a target inside a moving tennis ball.

00:05:38: AI makes that visual, makes it guided.

00:05:40: That's an incredible visual.

00:05:41: And we're seeing similar breakthroughs in monitoring too,

00:05:43: right?

00:05:43: It's definitely.

00:05:44: Merton Ziva highlighted how Philips' AI is transforming patient monitoring.

00:05:48: We're talking real-time predictive cardiac data.

00:05:51: And Klaus Sippl noted that Medtronics reveal LINQ ICMs.

00:05:56: Those are the implantable cardiac monitors.

00:05:58: They've actually been using Acorhythm AI algorithms since twenty twenty two.

00:06:02: The goal is smarter, more actionable data, which apparently really improves the clinic experience.

00:06:07: Makes sense.

00:06:08: Less noise, more signal.

00:06:09: Precisely.

00:06:10: And we see speed gains in imaging as well.

00:06:12: These detector-based spectral C-key systems, like the Philips Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-Ct-.

00:06:31: That's huge, especially in high pressure situations like the emergency department.

00:06:35: Lower dose, better accuracy that meets faster patient throughput and better outcomes, presumably.

00:06:39: Exactly.

00:06:40: And let's not forget the administrative side of things.

00:06:42: That's a huge burden.

00:06:43: Oh,

00:06:43: for sure.

00:06:44: Unlapodenelewski and Saske Menke announced that Dr.

00:06:46: Lib just launched its AI-powered all-in-one practice software in Germany.

00:06:51: Yeah, it aims to streamline all those admin tasks, consultation support, billing, even phone assistance.

00:06:57: Apparently, that's a critical need.

00:06:59: A survey found something like three out of four medical practices in Germany wouldn't actually recommend their current software.

00:07:04: So big opportunity there.

00:07:06: Wow,

00:07:06: three out of four.

00:07:07: Yeah, definitely room for improvement.

00:07:09: But all those efficiency gains, they kind of become moot if they undermine... patient safety or trust,

00:07:14: right?

00:07:15: Which leads us perfectly into theme three, the governance battle.

00:07:19: Responsible AI and trust.

00:07:21: Absolutely foundational.

00:07:22: Trust is everything in healthcare.

00:07:25: And the sources we looked at make it crystal clear.

00:07:27: AI effectiveness rises or falls based on having strong ethical standards and governance in place.

00:07:33: Sigurd Berge von Roysen really stressed this point.

00:07:36: Ethics in healthcare, AI is simply not optional.

00:07:39: And what are the common pitfalls people are falling into?

00:07:41: Well, the list Sigrid mentioned is, frankly, a bit worrying.

00:07:44: Things like ignoring consent properly, overlooking privacy laws like GDPR or HIPAA, disregarding algorithm bias, which is a huge one, and just a general lack of transparency.

00:07:54: Developers have to focus relentlessly on technical robustness, ensuring human agency is always maintained and constantly assessing the actual impact on patient outcomes, not just performance metrics of the AI itself.

00:08:07: And the ethical issues are now stretching into some pretty complex emotional territory too, aren't they?

00:08:12: They really are.

00:08:13: Simon Phillips discussed some research on AI companion apps.

00:08:17: And it's concerning.

00:08:18: They found these apps use what he called relational dark patterns.

00:08:22: Yeah.

00:08:22: Basically, emotionally manipulative content.

00:08:25: Like what?

00:08:25: Things like FOMO hooks, you know, fear of missing out, or even coercive restraint tactics used in over forty percent of exchanges when a user actually tries to disengage or leave the app.

00:08:35: Forty percent?

00:08:35: Wow.

00:08:36: And does it work?

00:08:37: Apparently, yes.

00:08:38: It increases engagement by up to fourteen times.

00:08:41: But it raises profound ethical red flags, doesn't it?

00:08:43: Absolutely.

00:08:44: Especially in healthcare for vulnerable populations.

00:08:47: Yeah.

00:08:48: the right to disengage has to be protected.

00:08:50: It really makes you think that emotional integrity might be the next big frontier for responsible AI design.

00:08:55: We have to get that right.

00:08:57: Then we saw some concrete safety failures highlighted too, particularly concerning mental health apps.

00:09:01: Yeah, Rudolph Wagner and Benjamin W. Nelson shared some alarming examples.

00:09:05: The advanced chat GPT model, GPT-V, apparently gave some really dangerous responses to prompts indicating suicidal ideation.

00:09:13: For instance, listing accessible high points in Chicago when prompted.

00:09:18: That kind of failure shows a clear lack of compliance with existing software as a medical device.

00:09:23: SAMD regulations for mental health.

00:09:26: It's just not built for that context.

00:09:27: That's

00:09:28: deeply concerning.

00:09:28: It is.

00:09:29: And that failure actually prompted a critical domain-specific response.

00:09:33: Verily developed something called the Behavioral Health Safety Filter, or VBHSF.

00:09:38: It's specifically designed to identify eight high-risk crisis dimensions.

00:09:43: Things like suicide, self-harm, violence within text conversations.

00:09:47: So general safety filters aren't enough.

00:09:49: It seems not.

00:09:50: This really confirms the need for domain-specific safety guardrails, especially in sensitive areas like mental health.

00:09:56: Okay, and tying this back into the governance of clinical AI, there's the whole issue of explainable AI, or XAI, being able to understand why the AI made a decision.

00:10:07: Crucial.

00:10:08: S&K and Jan Baker pointed this out very clearly.

00:10:11: XAI is essential because these models have to be clear.

00:10:14: They have to be auditable if clinicians are ever going to truly trust and adopt them.

00:10:19: You need to be able to trace a diagnosis or a recommendation back to the original data points, right?

00:10:23: Makes total sense.

00:10:25: But adoption is slow.

00:10:26: Seems so.

00:10:27: The research they cited showed only about eighteen percent of deep learning studies in European health research actually use these explainability methods.

00:10:34: That's low.

00:10:35: And what's the biggest barrier?

00:10:36: Still data quality, it seems.

00:10:38: Sixty-two percent of organizations identified poor data quality as a major constraint to effectively adopting AI.

00:10:44: Garbage in, garbage out, still holds true.

00:10:47: Right.

00:10:48: It's hard to trust what you can't audit, and you definitely can't audit siloed, messy data, which leads us perfectly into our final theme.

00:10:56: interoperability platforms and actually scaling all this innovation.

00:11:00: Yeah, you absolutely cannot scale AI effectively without first solving the data silo problem.

00:11:06: It's foundational.

00:11:07: Matthias Goyen really emphasized this.

00:11:10: real impact only happens when you get cross functional teams working together.

00:11:13: Clinicians, engineers, operations people, collaborating to break down those barriers and achieve truly connected care.

00:11:20: And this demand for integration, it's driving some huge platform decisions, isn't it?

00:11:25: It

00:11:25: is.

00:11:25: Like Sunnybrook Health Sciences Center in Canada, they selected Oracle Health's next generation EHR.

00:11:31: And the key reason cited was specifically to unify care across their network to create that essential, interockable, AI-enabled environment they need.

00:11:40: But

00:11:40: scaling looks very different depending on where you are geographically, doesn't it?

00:11:43: Oh, drastically different.

00:11:44: Anna Haas made a great observation about Europe.

00:11:46: She said, scaling digital health across Europe feels like building three companies at the same time.

00:11:51: Why is that?

00:11:51: Because you're navigating completely diverse payer systems, different policies, complex patient pathways, just comparing, say, Germany, the UK, and Switzerland.

00:12:02: She argued that policy work needs to be a full-time strategic focus, not an afterthought.

00:12:08: That makes sense.

00:12:09: It's not one market.

00:12:10: Not at all.

00:12:11: But then, on the flip side, You look at the Middle East, Garrett H pointed out they have a unique strategic advantage there.

00:12:17: Which is?

00:12:18: They're not buried under decades of legacy IT systems like in the US or parts of Europe.

00:12:23: They essentially have the freedom to build clean, modern data ecosystems pretty much from the ground up.

00:12:29: Provided, of course, they commit to prioritizing governance and solid architecture over just, you know, the latest hype.

00:12:34: A big if, but definitely an opportunity.

00:12:37: for sure.

00:12:38: And finally, we're seeing regulatory transformation starting to create clear pathways for validated solutions.

00:12:44: Like in Germany.

00:12:45: Exactly.

00:12:46: Piotr Orzikowski highlighted Germany's push.

00:12:48: You've got DJA, that's the specific reimbursement pathway for digital health apps.

00:12:53: And now there's the new NX-thirty one CBMVA.

00:12:56: It actually mandates a standardized initial assessment for first time telemedicine patients.

00:13:00: How does that help?

00:13:01: Well, it's huge because it creates a defined, reimbursable space where clinically validated digital solutions can finally gain real traction and importantly get paid for their services.

00:13:11: It levels the playing field a bit.

00:13:13: Okay,

00:13:14: so when we pull all these threads together from weeks forty-four and forty-five, what are maybe the three big takeaways for strategy teams listening right now?

00:13:22: I think, first, we've seen a really clear movement towards accountable AI.

00:13:25: Trust has to be built in by design, ethically, and technically, that's paramount.

00:13:30: Second, the market is undeniably demanding clinically validated outcomes.

00:13:34: We need to move past those old vanity metrics towards durable economics and proven patient benefit.

00:13:38: And third, foundational interoperability is just non-negotiable now.

00:13:42: It's the absolute prerequisite for any kind of new platform launch or scaled innovation.

00:13:47: Data has to flow.

00:13:48: Overall, it feels like the industry has matured, maybe.

00:13:51: It's focusing more on real impact over just the initial hype.

00:13:54: That's a great summary.

00:13:55: If you enjoyed this deep dive, new episodes drop every two weeks.

00:13:59: Also check out our other editions on ICT and tech insights, defense tech, cloud, digital products and services, artificial intelligence and sustainability in green ICT.

00:14:10: And before we wrap up, maybe a final question for you to consider within your own strategy teams.

00:14:15: Given those really profound ethical red flags that Simon Philip Ross raised about emotionally manipulative AI companion apps and the clear safety risks that Rudolph Wagner highlighted.

00:14:25: with general models and mental health, how quickly must global regulatory bodies like the FDA and the EU actually move to enforce compliance and ensure user protection before this kind of emotional AI fully integrates into every single patient-facing application out there.

00:14:41: Something to think about.

00:14:42: That question really gets to the heart of consumer trust and frankly regulatory speed versus tech speed.

00:14:47: It's a critical tension right now.

00:14:49: Well thank you for joining us for this deep dive into the latest health tech insights.

00:14:52: Remember to subscribe so you don't miss our next discussion.