Vinay Bhargava

Great article from Jessica Lamb and Eric Larsen on the potential for AI in healthcare, and why this time is different... https://lnkd.in/eYCvt8J7

14

Because you can mandate all clinicians to use the same AI scribe, we’ve fallen into a dangerous trap: believing we can force patients to engage with a one-size-fits-all platform. It’s time we stopped treating Digital Health like staff software and started treating it like what it actually is: a Clinical Intervention. There’s a strange obsession in Health Tech with the idea that there should be ONE solution for everything - but that limits our ability to achieve the best patient outcomes. In medicine, we recognize that different patients respond to different interventions - and part of the job of clinicians is to find an intervention that patients can adhere to and is effective. Take hypertension: a younger patient might be more committed to try lifestyle changes first - whereas, an elderly patient with mobility issues may be more adherent to medication. Or pain management: some patients find medications more effective, others may need injections, while others find a way to manage pain through physical therapy. Yet in Health Tech, we often avoid personalized approaches to engaging patients - at the cost of patient adoption and outcomes: → “We should just have a text-based AI agent navigate a patient through patient care journeys (e.g. pre/post-surgery)” when all the data shows that a multi-channel approach of text, email, mobile app, web app and patient portal integrated experiences have the best patient adoption and outcomes. (I actually do think conversational AI will play a part in this… just no one has proven it with real clinical evidence yet.) → “We should have a patient portal only approach to engaging patients” when all the data shows patient access and engagement is maximized by integrating the patient portal with purpose-built, third party care journey platforms (like SeamlessMD). Why is it that in medicine we recognize that there often aren’t one-size fits all interventions, yet in Tech, we often look for the one perfect solution? Sure, I can appreciate the usual answer: reducing IT overhead through vendor consolidation, desiring a single user experience etc. - it’s all fair and rational. But we need to look at internal staff differently from patients. You want every staff member using Microsoft Teams? Fine. You want every clinician using the same AI scribe? Fine. But you expect every patient to only engage through the patient portal? Or through an SMS-based AI agent? Good luck with that. The truth is that patient journeys are inherently complex - you’re dealing with a huge diversity of patient demographics (SES, health literacy, age, etc.) and medical complexities. Technology leaders need to think of Digital Health tools less like Microsoft Office and more like medical interventions - where one-size fits all interventions often don’t exist. Otherwise, we will keep trying to search for the “perfect” Tech solution, and keep asking why half the patients don’t use it, and why clinical outcomes don’t look any different.

162

39 Comments

Really enjoyed Shiv (Shivdev) Rao’s conversation with The Information TV on where Abridge is headed and what it signals for AI in healthcare. What started as a way to reduce “pajama time” for clinicians is becoming something much bigger: a platform built around the conversation between patient and clinician. What stood out most was Shiv’s framing of the opportunity: using AI not just to save time, but to save money and improve care. https://lnkd.in/gr2T7sgS

2

What happens when you put humans not just technology at the center of AI in health care? I had the privilege of joining Shikha Jain, MD, FACP, on the Healio Oncology Overdrive podcast to explore how AI is reshaping oncology and what it will take to do this responsibly. Here's a sneak peek at what we covered: ✅ How to educate patients and clinicians to use AI safely — especially in marginalized communities ✅ A look at how CROSS Global Research and Strategy, LLC, partners with organizations to apply AI in practice ✅ Tackling concerns around trust, bias, and equity in clinical AI ✅ Why data governance + prompt design must go hand in hand ✅ What success in this space could look like over the next 10 years — and how to navigate today’s climate while building toward it I'll leave you with this hot take: AI in oncology isn’t about replacing clinicians. It’s about equipping them with tools that are evidence-based, transparent, and designed to amplify human expertise. 👉 Listen to the full conversation here:https://lnkd.in/ebfxkmuf #ArtificialIntelligence #AIinHealthcare, #Oncology, #Cancer, #Innovation

8

1 Comment

As HLTH Inc. 𝗸𝗶𝗰𝗸𝘀 𝗼𝗳𝗳, I’ve been thinking back to the incredible innovators we hosted at the 𝟮𝟬𝟮𝟱 BioSTL Global 𝗛𝗲𝗮𝗹𝘁𝗵 𝗜𝗻𝗻𝗼𝘃𝗮𝘁𝗶𝗼𝗻 𝗦𝘂𝗺𝗺𝗶𝘁 - nearly 40 companies tackling some of the biggest challenges in care delivery and AI-enabled access. A few that stood out pushing the frontlines of how care is delivered:  • Hippocratic AI – Scaling safe, multilingual patient engagement at record volume.  • Sayvant – Ambient documentation purpose-built for emergency and urgent care.  • Pomelo Care – Virtual maternity model cutting NICU admissions by 58%.  • Vitalic Health – Geriatric psychiatry improving PHQ scores by 42%.  • Isaac Health – Virtual memory clinic supporting early dementia diagnosis.  • 𝗔𝗽𝗼𝗴𝗲𝗲𝗖𝗮𝗿𝗲 – Embedded ED care teams delivering 13× ROI.  • Pulsenmore – At-home ultrasound connecting mothers with remote OB support.  • 𝗕𝗿𝗶𝗱𝗴𝗲 𝗦𝗼𝗰𝗶𝗮𝗹 – Tech-enabled social support for older adults. Exciting to see so many tackling workforce strain, maternal health, and access — themes sure to be front and center at HLTH this week.

44

Last week a16z published "Infinite Healthcare: What's It Worth?" (Jay Rughani, Jane Rhee, Julie Yoo). They're right about almost everything: AI ends clinician scarcity, the information layer of medicine deflates toward the price of electricity, metering dies, access pricing wins, the mix shifts from reactive to proactive. But the essay asks every question except one: Who owns it? When the meter dies and care becomes a subscription, the subscription becomes the new meter — a per-member toll on a service whose marginal cost is collapsing toward zero. The essay names the destination plainly: "the companies that price for abundance will capture the biggest prize." The companies. We've run this experiment — rides, rooms, music, gig work. Act one, everything's a deal. Act nine, the take rate has matured and your ratings have a landlord. Healthcare is about to re-run it on the most intimate labor in human life. Their four pricing models — per task, per workflow, per episode, per patient — map act one. There's a fifth: Per member — where the members own the vendor. Three things the essay leaves out: The actual workforce. The majority of American care labor isn't "clinicians aided by AI agents" — it's 53 million unpaid family caregivers. Consumer-directed Medicaid already pays family caregivers in every state, through intermediaries taking 40–50¢ of each dollar. A cooperative takes 8–15 and returns the rest as wages and equity. Jevons doesn't just apply to inference. It applies to presence. The governor. "More healthcare is bad" is half-irrational — the other half is overdiagnosis and engagement loops. An access-priced vendor's incentive is engagement. The durable alignment layer is ownership: consumption steered by members who vote the protocols and harvest the savings. The ending. Nobody in the essay grows old or dies — Jill just "retires on her own terms." But a quarter of Medicare is spent in the last year of life, mostly in places people swore they never wanted to be. A subscription doesn't sit with you at 3 a.m. A member-owned network — neighbors on a time-banked ledger, a granddaughter as the paid caregiver of record — does. Modeled across one family's final decade, the owned path preserves $600–900k per generation. Not conjured — un-destroyed. Price for abundance, yes. Then finish the thought. Price × Quantity is only frightening when the expense leaves the community. When members own the vendor, the quantity is care your family gives and receives — and the margin is your wage, your dividend, your mother's house staying your mother's house. The fifth model is already running in Colorado: found your family's care co-op in 10 minutes, keep five health promises, accrue a floor of care-hours, own the rails. Open-source. Federated. Every company in America is born in a filing cabinet. Yours can be born at a table. → co-op.care/found

6

4 Comments