Sigrid Berge van Rooijen’s Post

This is such an important breakdown. 🚀 As leaders, we often view AI in healthcare as a technology problem — but the true barriers lie in mindset, systems, and trust. Integration, regulation, ethics, and data quality are not obstacles to fear, but opportunities to reimagine how we deliver care. Driving meaningful change requires: 🔹 Courage to challenge legacy systems while investing in scalable infrastructure. 🔹 Collaboration across disciplines to align regulation, innovation, and patient safety. 🔹 Commitment to transparency and trust, ensuring AI enhances — not replaces — the human connection in healthcare. Unlocking AI’s full potential isn’t just about fixing gaps; it’s about leading with vision, adaptability, and accountability. Those who embrace this change mindset will help shape a more equitable, efficient, and human-centered future of healthcare.

We will struggle to find the sweet-spot between reckless adoption and overcautious resistance to AI use in Healthcare. Regulation especially struggles to keep pace with changes, and attempts to suppress its use risks people going off-piste either deliberately or naively. A bumpy ride ahead I fear but we must move forward and manage associated risks. It will be worth it.

Anybody can build an AI solution. Anybody can get access to a large language model. It’s commoditized at this point. What you train that on—the data you give it access to—is really the secret sauce.” – Dr. Mitesh Rao, MD, CEO & Co-Founder, OMNY Health. AI is revolutionizing healthcare and pharma, but success doesn’t come from algorithms alone. As Dr. Rao explains in our latest Q&A, the real differentiator is high-quality, representative, and comprehensive data. From accelerating drug discovery to improving patient care, the future of AI in healthcare will depend on: ✅ Access to unstructured clinical data that captures the why behind patient journeys ✅ Investments in regulatory-grade, de-identified datasets ✅ A shift toward data democratization so innovation isn’t limited to a few players OMNY Health is helping bridge the gap by providing one of the nation’s largest networks of de-identified clinical data, empowering Pharma, MedTech, and healthcare systems to build reliable, impactful AI solutions. Read the full Q&A with Dr. Rao to learn why data quality, not just data volume, will make or break AI in healthcare. #AIinHealthcare #DataQuality #PharmaInnovation #DigitalHealth #HealthTech #OMNYHealth #DrugDiscovery #FutureOfHealthcare

🗽 AI in MedComms: Beyond Hype, Into Workflows AI has become the most overused word on LinkedIn in the past year — often framed as an oracle or a threat, but rarely discussed in terms of how it actually helps people do their jobs. In Medical Affairs and MedComms, the stakes are higher than most sectors: compliance, evidence integrity, and stakeholder trust are non-negotiable. Which is why the real question isn’t “can AI do magic?” but: 👉 How much time are teams still wasting on tasks that could be accelerated without compromising compliance or quality? --- Why sAÏmone exists That’s the problem I built sAÏmone to address. Not as another generic chatbot or “AI experiment,” but as an audit-ready, compliance-aware assistant built specifically for boutique MedComms agencies and small biotech teams. Where generic AI tools stumble on hallucinations or lack of traceability, sAÏmone is designed to embed governance and auditability from day one. --- Practical benefits for teams 1. Cut wasted time on structuring Turning rough notes into a structured publication plan, drafting stakeholder maps, or shaping tactical timelines — sAÏmone accelerates the repetitive prep work that eats senior bandwidth. 2. Faster evidence navigation Instead of spending hours digging through PubMed, EMA/FDA guidance, or HTA assessments, sAÏmone surfaces relevant insights with citations and traceability. That’s time reclaimed for higher-value thinking. 3. Audit-friendly outputs by default Every plan, recommendation, or summary is logged, cited, and designed to withstand internal review and external scrutiny. Outputs aren’t “black box,” they’re built to be audit-ready. 4. Amplification, not replacement sAÏmone isn’t a “magic box.” It doesn’t replace human expertise, it amplifies it — giving agencies and biotech teams more bandwidth to focus on science, strategy, and value. --- What this means in practice Boutique agencies can take on bigger clients and projects without ballooning headcount. Small biotechs can operate with the strategic discipline of much larger organizations. Teams spend less time wrestling with admin and more time delivering insight, value, and impact. --- Closing thought AI won’t replace Medical Affairs or MedComms expertise. But those who integrate it well will move faster, deliver more consistently, and build more trust. 🗽 That’s the torch sAÏmone was created to hold. #MedicalAffairs #MedComms #LaunchExcellence #PharmaAI #BoutiqueAgencies #DigitalTransformation #Compliance #AuditReady #LifeSciences #Biotech

Closing the Predictive AI Divide in Healthcare: How Paradigm and Krista Can Help By Vin Matozzo, CEO, Paradigm Venture Group The Opportunity — and the Divide More than 70% of U.S. hospitals now use predictive AI, a five-point jump from last year, according to the Assistant Secretary for Technology Policy and ONC. But adoption isn’t equal. System hospitals report 86% usage, versus only 37% of independents and 56% of rural hospitals. That gap means smaller facilities risk falling further behind in identifying high-risk patients, optimizing staffing, and controlling costs. Why Adoption Stalls Lagging hospitals face common challenges: Limited resources — no dedicated analytics teams. Governance burdens — accuracy and bias checks require cross-departmental oversight. Integration costs — EHR and billing connections often run six figures without IT depth. The result is uneven adoption and widening disparities. Enter Krista: Orchestrating Predictive AI Without the Overhead Krista isn’t another AI model — it’s a natural language orchestration layer that lets hospitals manage AI tools through plain-English dialogue. Simplified Governance: Automates monitoring for accuracy and bias while keeping clinicians in the loop. Faster Integration: Connects predictive models to EHR, billing, and scheduling without heavy custom work. Affordable Entry Point: Reduces IT lift and labor costs so smaller hospitals can benefit. Practical Use Cases Three immediate Krista-enabled wins: Revenue Cycle: Predict denials and improve cash capture. Patient Throughput: Optimize scheduling and escalate issues automatically. Population Health: Flag high-risk outpatients and ensure follow-ups. Closing Thoughts The ASTP report is clear: predictive AI is no longer optional, but adoption remains uneven. At Paradigm, we believe AI orchestration — not just algorithms — is the key to closing this divide. Krista enables hospitals of any size to adopt and scale predictive AI safely and affordably. It’s time to bring predictive AI to every hospital — not just the ones with the deepest pockets.

Medical AI must be explainable—not a black box Opaque model logic can undermine practitioner trust and reduce adoption. Explainability is more than a nice-to-have—it’s foundational. Redwerk illustrates how clarity improves clinician confidence and outcomes. #ExplainableAI #AITransparency #ClinicalAI https://lnkd.in/gWrivv-j

💡Health care systems adopting AI without clear strategy 🔎Around 2/3rd of health care executives surveyed by Sage Growth Partners are investing in AI for administrative operations and clinical care, but only 13% of executives have a defined strategy for clinical AI use. 📍 Only 12% of survey respondents said that AI is reliable, and around half said appropriate use of AI is a challenge. 📝Read more 👉🏽 https://lnkd.in/gcMPJDGv ... and reach out to our EPAM Systems and EPAM Continuum Teams if you would like to partner on implementing AI with ROI.

Just spent 20 years building AI systems that failed spectacularly, then succeeded beyond imagination. The pattern in healthcare nobody talks about: Pure AI automation = dead end Human-AI hybrid teams = exponential results Here's the nuance everyone misses: Healthcare AI isn't failing because the technology isn't ready. It's failing because we're asking the wrong question. Instead of "How do we replace doctors?" the winning companies ask "How do we make doctors superhuman?" Think of it like Formula 1. You don't remove the driver and let the car drive itself. You give the best driver the most advanced car and watch them do impossible things. The companies scaling aren't deploying chatbots. They're creating hybrid intelligence systems where AI handles pattern recognition at superhuman speed while humans handle judgment calls requiring empathy and ethics. The stolen playbook smart health systems are running: First, they identify tasks doctors hate doing (documentation, routine analysis) - not to cut costs but to free up cognitive bandwidth Then they deploy AI as a research assistant, not decision maker - which seems limiting until you realize doctors with AI assistants are 10x more accurate Finally, they measure doctor satisfaction, not just efficiency - and that's when patient outcomes explode Hot take: In 18 months, asking "Will AI replace doctors?" will sound as outdated as asking "Will calculators replace mathematicians?" The future isn't human vs. AI. It's enhanced humans vs. regular humans. Who's building hybrid teams? #HealthTech #AITransformation #FutureOfHealthcare #DigitalHealth #HealthcareInnovation

We are hard at work measuring the effectiveness of our AI copilot for clinicians, Hyperscribe. Sharing some observations on the three most common approaches to evals I see today: 0️⃣ "No-op" — for all cases, do nothing. Make performance claims with no credible evidence, and target buyers who don't question it. 1️⃣ Comparative labeling — for a given case, physicians sort two or more notes from worst to best. 2️⃣ Case-specific rubrics — for a given case, physicians develop a structured set of weighted criteria to score any note for the case. OpenAI's HealthBench is almost surely the largest example of this approach. No-op has no place at Canvas Medical. Comparative labeling provides a high quality feedback signal, but it's prohibitively expensive for anything other than small scale one-off studies. No continuous, production-grade governance system can rely on comparative labeling. Rubrics are scalable — even the case-specific version. They can be used beyond one-off studies and are suitable for always-on AI governance systems. But we don't know much about the quality of the signal: do they produce scores that actually reflect a reliable ground truth that would agree with the results of comparative labeling? Our team has been working on a rubric validation methodology that we hope will combine the best of both worlds — the solid ground truth of comparative labeling with the scalability of rubrics. Here's the basics on how it works: 1. For a given case, we AI-generate an initial case-specific rubric with detailed scoring criteria. 2. We have Hyperscribe produce several draft notes under different parameter settings. 3. A vetted clinician then orients to the case and labels the best and worst note in a comparative step. 4. The clinician then iteratively edits the rubric until it reliably scores the “best” note higher than the “worst” note. We repeat the entire process independently with a second vetted clinician, yielding two validated rubrics per case. The validated rubrics are ideal for sustained development and experimentation, as well as ongoing surveillance of deployed agents. If you saw Lenny Rachitsky's post about "CC/CD" — that's what this unlocks (link to post in comments). Want to learn more? Just comment and I'll share what I can. Or wait for the paper!