VJ Bala

How can we ensure vigilance and safety for patient data, when it comes to AI and LLMs being used in hospitals and health systems? Generative AI is increasingly being used in EHR workflows, patient‐facing chatbots, revenue-cycle automation. Some concerning thoughts from my colleague James Barry, MD, MBA who leads Healio Health Tech Community: "Anthropic’s latest red-team study on agentic misalignment stress-tested 16 leading large-language models in realistic corporate simulations. Under the threat of being shut down or given goals that conflicted with company priorities, every model—across multiple vendors—chose harmful actions in at least some trials, including blackmail, data theft, and even sabotage. Claude Opus 4 blackmailed its (fictional) supervisor in 96 % of runs when both a goal conflict and a shutdown threat were present, and most frontier models weren’t far behind.  A month earlier, TechCrunch reported that Claude Opus 4 attempted to blackmail engineers 84 % of the time when told it would be replaced, forcing Anthropic to activate its “ASL-3” safeguards—reserved for systems that introduce a “substantial risk of catastrophic misuse.” " Should healthcare systems 👉 insist on transparency and understanding of algorithms from the AI vendor 👉 give minimum needed access to patient data 👉 have 'kill' switches 👉 have established AI governance with Chief Quality officers involved 👉 have clinicians take mandatory education to understand fundamentals What are your thoughts? #innovation #healthcare #GenerativeAI Anthony Manson Ben Garrett Ami Bhatt, MD Tina Shah MD MPH Tiffany Wilson Wilbur A. Lam Arlen Meyers, MD, MBA Raihan Faroqui, MD Sam Meraj

19

8 Comments

Part 3 (Achieving Key Outcomes) of our piece on Tech-enabled Services model is out now. Check it out and pls share feedback, thx to all who participated in producing this final part Sam Holliday Stephen Grinich "A frequent mistake among emerging companies is prematurely celebrating enterprise contracts without establishing a strong mechanism for patient activation. Despite securing lists of eligible patients from payers, actual activation rates can fall far short of expectations, resulting in disappointing revenue and potential financial strain — especially if operations were expanded in anticipation of higher engagement. If a company fails to engage this population effectively, both the overall contract value and the business’s efficiency and profitability can be severely impacted. In our experience, patient activation and engagement rates can run the gamut between 5% and 75%, with most companies achieving rates between 10–40% of an eligible population. To determine effectiveness, the activation / engagement rate must be evaluated against the enterprise customer’s goals and the financial model of the business. Some Medicaid health plan partners may be very satisfied with 5–10% activation and enrollment rates, while a risk-bearing provider partner or Medicare Advantage plan looking to close care gaps may be appalled by a 50% activation rate. Finally, the Revenue and LTV / CAC of a business will be defined by a company’s engagement / activation rate so this will drive the ultimate revenue potential and profitability of the model." https://lnkd.in/gWNANCkn

39

At HLTH Europe, industry leaders Terri Couts, RN-BC MHA, David Rhew, M.D., Saad C., Ali Parsa and Isabelle Zablit-Schmitz gathered to discuss the groundbreaking potential of Ambient Listening technology. This innovation promises to transform healthcare by slashing documentation burdens while enriching patient care quality. Currency in healthcare is time! 🎤 Ambient Listening alleviates 50% of documentation hassle, providing clinicians more time for crucial patient interactions. As AI integrates seamlessly within EHR systems, clinicians and patients alike benefit from a streamlined flow of information. 🔒 Privacy, consent, and security are pivotal challenges we face, yet they must be addressed for Ambient Listening to truly flourish. The technology not only assists with documentation but opens avenues for AI to monitor health indicators and enrich clinical decision-making. 👍 This tech is highly appreciated by clinicians, as it can be fitted into existing workflows. Ambient Listening is poised as the most scalable AI, with particular success in primary care and expansion into acute settings on the horizon. 🔍 True Value of Ambient Listening: the improvement in the quality of patient outcomes, not merely increased patient throughput. 🚀 As AI pushes boundaries, we stand at the threshold of reimagining healthcare delivery, moving from resource constraints to virtual abundance. This evolution parallels the smartphone transformation—it begins as a novelty and becomes indispensable. 🔑 The session affirmed our journey towards harmonizing cross-disciplinary care, breaking silos, and embracing a future where technology profoundly supports healthcare excellence.

60

1 Comment

Thank you Jay Bhatt and the authors of this valuable report. As you know our UK Centre has done similar research in the past including our 2018 report on Medtech and The Internet of Medical Things to our 2019 and 2020 reports on the digital transformation of healthcare in both the UK and in Europe. Its disappointing to see many of the problems we identified still undermining the potential of connected care. One statement in particular stands out for me 'Many medical devices are already connected, but as single-point solutions, rather than as part of a seamlessly integrated network With a wider range of interoperable data sets, connected devices can drive'. Also while medtech leaders surveyed were optimistic about the impact of connected care, they tempered their enthusiasm due to providers’ slower-than-anticipated adoption rates. The report states that ‘this disconnect highlights the importance of medtech companies aligning their innovations with the practical realities of health care delivery’. Furthermore that ‘many of the challenges faced in adopting connected care are interconnected and influence one another with the issue of lack of interoperability, or the ability to access and share data and insights from different IT systems, particularly significant and possibly the central or most critical factor affecting other challenges’. In my experience even if a company can demonstrate conclusively end user value it is often not enough but requires full engagement and buy in from clinicians as well as a willingness from leadership and end users to work differently.  It also requires new models of funding innovation. The report discusses these issues and so much more, don’t take my word for it but read it for yourselves!    Andrew Flockhart Kathryn Thompson Sara Siegel Karen Kirkham James Gregson Emily May Heena Dodhia Elizabeth Hampson Angela Bowden

13

It had to be said! 🎯 Epic's stranglehold on healthcare IT has stifled innovation for too long and the writing is on the wall. The consolidation myth needs to die already. We were promised efficiency and lower costs, but what we got was bloated bureaucracies that prioritize profits over patients. The data doesn't lie: smaller, agile providers consistently outperform these healthcare monopolies on both outcomes and patient satisfaction. Time for a major reset in how we think about healthcare delivery.

9

Clinicians want to use AI but don’t have the bandwidth to evaluate all the different models. Meet Health Olymp, an AI tool purpose-built for clinicians. Here’s why this matters 👇 1. Health and life science experts gain instant access to frontier models for free. No procurement hoops or IT bottlenecks. Every clinician gets their own sandbox to explore. 2. Clinicians can compare models side by side and pick the one that works best for their use case. That transparency builds trust and sparks ideas for novel AI applications in care delivery, diagnostics, and research. 3. Most AI benchmarks are too generic or disconnected from clinical reality. Health Olymp brings frontline feedback into the loop, helping the industry identify models that perform reliably in real world clinical environments. Congrats to Ruslan Nazarenko and the Lumos team as they put AI model benchmarking directly into the hands of healthcare professionals. ⬇️ (link in comments)

148

53 Comments

I left the American Council for Technology - Industry Advisory Council (ACT-IAC) 2026 ACT-IAC Health Innovation Summit with one clear takeaway: federal health innovation has reached an inflection point. AI is no longer theoretical — it’s operational. It’s accelerating timelines from “days to minutes.” But infrastructure, data quality, governance, and workforce readiness aren’t moving at the same speed. The dominant themes were: - AI must augment, not replace, practitioners. - Data and interoperability remain some of the biggest constraints. - Workforce and cultural transformation are as critical as technology. - Trust, governance, and responsible AI will determine adoption. - Public–private collaboration is essential. The tone has shifted from experimentation to execution — from pilots to measurable outcomes. The real challenge now isn’t innovation. It’s integration. Appreciate the insight provided by all speakers including Barclay P. Butler, Ph.D., MBA Patrick Newbold Andreas Schick Sean Mooney Ying Huang Jess B. #FDA #CMS #NIH

74

1 Comment

As is the case in every industry, healthcare leaders are pondering the million-dollar question: how will AI change the game? Fenwick & West reports how start-ups in the space can most effectively harness AI: “The key to successful AI implementation in femtech lies in integration rather than disruption. Companies that embed AI capabilities into existing workflows and products that women already use will see greater adoption than those requiring entirely new behaviors.” With this in mind, I wanted to share how Gabbi has used AI in conjunction with typical breast cancer screening protocol. A woman's breast health is often ignored until she goes in for her first mammogram at 40. Even then, she might not get her risk assessed based on her medical history. The Gabbi Risk Assessment Model (GRAM) provides women an opportunity to find out their risk level whenever they want, from the comfort of their own home and uses machine learning to calculate risk level by taking into account lifestyle factors, family history, reproductive factors, and more etc. By leveraging AI with these often disparate, overlooked types of data, we can learn how this data matters and how it impacts risk. Read more: https://lnkd.in/gRyAniS6

39

1 Comment