Alex Oshmyansky, MD, PhD

As the discourse shifts from models to compound AI systems / agents, we need better AI benchmarks to evaluate multi-modal and multi-step task performance, especially in healthcare and life sciences. When we wrote the first paper demonstrating ChatGPT's performance on the USMLE, we chose the US Medical Licensing Exam as a benchmark for accessibility, speed, and ease. This benchmark was never intended to represent AI model performance on real-world clinical tasks. Today, I still see so many research teams and startups using benchmarks (like the USMLE) that are ill-fitted for assessing the true clinical or scientific performance and utility of the models they are developing for real-world contexts. Benchmark development may be seen as a "less sexy" area of research, but it is of paramount importance. Years after the rise of the transformer, we still lack adequate benchmarks for so many single-step tasks in biomedicine. With compound AI systems (i.e., architectures that integrate multiple AI models to perform complex tasks) emerging, we need new benchmarks for agentic behaviors. I'd argue that developing an agent with novel capabilities without at least proposing a companion benchmark (if an industry standard does not yet exist) may hinder the adoption of said agent, especially for high-stakes workflows. Designing more benchmarks that capture/simulate real-world clinical and scientific workflows will help us mitigate the major discrepancies observed between in silico and in vivo performance and better support safe + effective deployment of AI in biomedicine. There are already brilliant people focused here, and we need more. DMs are open if you're researching or working in this area of multi-step/multi-modal benchmarking in healthcare and life sciences! #healthcare #ai #artificialintelligence #generativeai

100

18 Comments

⚜️I had the honor of being an invited speaker at this week’s MedAxiom - An ACC Company #CVTransforum meeting in Atlanta, where I was invited to talk about my biggest passion: use of #artificialintelligence to optimize quality and delivery of clinical care. ❗️The AI healthcare market's projected growth from $20 billion presently to as much as $200-$400 billion by 2032 underscores the urgency of our understanding and strategic adoption of this transformative technology. 🔹Cardiovascular 🫀imaging is the subfield of Cardiology with the broadest embracing of #ai and for good reasons: ✅ automated image analysis ✅ automated detection of abnormalities ✅ automated prognosis and planning of next steps in treatment. 🔸AI-enabled cardiovascular interventions are already here and they hold the promise to optimize procedural planning and execution. 🔹The rise of #telehealth and "hospital at home" models, catalyzed by AI-driven remote monitoring and virtual care, promises to redefine access and convenience for our patients, much like Amazon did for retail. 🔸Large language models #llm and their potential to streamline clinical documentation and decision support through seamless human-AI dialogue is truly awe-inspiring. 🔹AI-based clinical decision support algorithms are already in clinical use in a number of hospitals and could profoundly elevate diagnostic accuracy, treatment selection, and risk stratification by synthesizing multidimensional data and evidence-based guidelines. While extremely promising, I also highlighted some key caveats and limitations of AI in healthcare. ⚠️ Data quality, bias, privacy concerns, regulatory hurdles, and the need for robust validation. 💚 AI should be viewed as a powerful tool to augment human expertise rather than replace clinicians entirely. 🙏🏻 Grateful to ACC MedAxiom for hosting this exceptional meeting and look forward to continued dialogue on advancing the adoption of AI for the betterment of cardiovascular health globally. #innovation #healthcare #healthcareinnovation #healthcarequality #medicine #cardiology American College of Cardiology Ami Bhatt, MD Jerry Blackwell, MD, MBA, FACC Guy David Lori Walsh Aashima Gupta Fatima Paruk, MD, MPH Jennifer Joe, MD Jennifer N. Avari Silva Mpirik (Acquired by Tempus AI) Tempus AI Asif Ali, M.D. Lisa Crymes Pierre Elias, MD Ashley Beecy MD, FACC Shilpa Patel Aidoc University of Washington University of Alabama at Birmingham Harvard University John Michael Maier Pat Lara Ana Mercurio-Pinto

56

3 Comments

“Cardiology continues to have the second largest number of clinical artificial intelligence (AI) algorithms cleared by the U.S. Food and Drug Administration, trailing only radiology. The FDA's latest update covers all AI approvals through May 2024. There are now 122 FDA-cleared AI algorithms specific to cardiology. This makes up 14% of the 882 clinical AI algorithms now on the U.S. market. The new FDA numbers include market clearances up to March 31, 2024. The first approval of a clinical AI algorithm was in 1995. The first cardiology specific algorithm was for HeartFlow FFR-CT in 2016. That algorithm was also the first AI to be included in a cardiology guideline when the 2021 AHA/ACC/ASE/CHEST/SAEM/SCCT/SCMR Guideline for the Evaluation and Diagnosis of Chest Pain was released.” MedTech Innovator American Heart Association #epeeps Heart Rhythm Society American College of Cardiology LSI Capital Factory Company Ventures MedTech Strategist Health Wildcatters Techstars

49

Advancing healthcare through tech requires playing the long-game in order to access and navigate the needs of the myriad of stakeholders: physicians, patients, payers, regulatory, IT, work-flow infrastructure/operations and more. Excited to join this important panel discussing The Roadmap to Integrate AI into the Healthcare Systems. Congrats to #BentleyUniversity for leading the conversation.

2

Takeaways from week two Stanford Med216: Generative AI and Medicine - “Be part of the wave of innovation, don’t let it crash on you” with Dr. John Hennessy - Alphabet Chairman and former President Stanford University. Pulling out a key thread from Dr. Hennessy’s session was his take on technology coming in waves of innovation: • We are in the AI wave of innovation - we have seen previous waves of innovation with the internet and with mobile. This is the wave to learn about, contribute to, and be a part of. • What’s different with AI innovation? - The speed of adoption and rate of change in this wave is unprecedented. Previously, we have seen technology take years for adoption and for its impact to be seen. We are going to see changes much faster with AI and need to react specifically around impacts to education and jobs.  • Role of education - Undergrad Education – Focusing on creating lifelong learners - Graduate Education - learning enough to be fearless to push forward when a new area comes along. These goals are not different in the AI wave of innovation - now we just have more people studying linear algebra.   • Regulatory environment is very tricky - the technology can be used for good or bad, and just because it can be used for bad does not mean we outright ban it. There are no easy answers to the copyright/usage of content. If we over-regulate we will push innovation elsewhere.  • What’s coming down the pipe?  We aren’t close to AGI. The improvements seen in the past few years have certainly shifted the timeline closer than most thought, but we are still years away. The Transformer architecture was a big step forward, but this isn't the end of innovation, we will continue to improve how we are building AI. What’s exciting in healthcare? • Patient interfacing - explaining medical concepts and treatments in ways that patients can fully understand. • Diagnostic error reduction – AI should be able to reduce errors in diagnosis and reduce the time it takes to reach the correct diagnosis. • Research and drug discovery will be made much more efficient with AI. • Overall with AI patients will take a larger role in their own healthcare journey. Quick hit questions: •Contrarian truth - "Individuals can change the world." •Key advice - "Remain intellectually curious for your entire life." Thanks to Longsha Liu and Bryant Lin my co-instructors for making this course possible. #genai #lifelonglearning #futureofhealth #AI (image created with Dalle)

79

4 Comments

The role of AI and new technology to aid cardiology “As cardiology starts to face staffing shortages and continued decreases in CMS payments, Biga believes technology advances need to be leveraged in cardiology to help improve outcomes, make workflows more efficient to do more with less, and to help cut costs.” “Many of the solutions that can help were displayed on the expo floor, including telehealth and real-time remote monitoring systems. "I think this area will experience an explosion in use," Biga said. Another major area that will likely see an explosion of use is artificial intelligence (AI). She said AI will begin playing a rapidly growing role in cardiology in the coming years to augment cardiologists and other clinical staff. And not just on the front lines of patient care, but also in the back offices where AI will play a growing role in IT systems, smoothing workflow, analytics, automated quality metrics capture and many other areas. "We must use AI to reduce administrative burdens and improve patient outcomes," Biga explained. “

17

1 Comment

Here's another fundraising roundup spotted in the healthcare wilds 👀 🍃 ✅ Sword Health raised $130M (new valuation: $3B). ✅ Better Health raised $14M to deploy expanded payer partnerships. ✅ Alzheon, Inc. | Preserving Future Memories raised $100M Series E to advance development & commercialization of its Alzheimer’s treatment. ✅ CinRx Pharma raised $73M to continue its ‘Hub and Spoke’ model of startup formation. ✅ Eko Health raised $41M to scale AI-driven heart and lung disease detection. ✅ Anterior (formerly Co:Helm) secured a $20M Series A to unlock admin efficiencies for healthcare payers. ✅ SamaCare raised a $17M Series B to scale specialty drug prior authorization platform. ✅ Plenful secured a $17M Series A to streamline pharmacy operations with AI-powered workflow automation. ✅ Grayce raised a $10.4M Series A led by Maveron amid a caregiving crisis. ✅ Stitch Employment Services Inc PEO raised $8.75M in seed to support independent medical practices. ✅ Posterity Health secured $6M to fund expansion and technology acceleration. ✅ Sware raised $6M to accelerate life sciences innovation by automating FDA-mandated software validation. ✅ Keragon raised $3M, as announced in @Conno Christou’s LinkedIn post. As always, if I missed you, let me know! Each week I'm spotlighting an innovative healthcare startup. I just ask for a bit more transparency. All submissions are confidential; submit yours here 👇 https://lnkd.in/gF4DrAvJ Looking for more fundraising roundups, alongside healthcare industry news and analysis? Subscribe to Hospitalogy, my 2x/week newsletter. Join 33K+ other investors, executives, and healthcare professionals: https://lnkd.in/g-nfRqW6

39

3 Comments

LSI, BioscienceLA, #LAMedtechweek, and HLTH's #VIVE2024 event resulted in the latest GTM Pulse Check offering which has successfully taken off! As a result, I am continuing on last week's series of go-to-market tactics. Crafting a robust marketing plan is essential for medical device companies to thrive in today's world. By leveraging these outlined sections, organizations can optimize their marketing strategies and drive meaningful impact.This post takes a step back to outline the sections of a marketing plan. Read on 👇🏼 Looking for support in creating your marketing plan?  Let’s chat! Book here 👉🏼 https://lnkd.in/g92YA9WK _____ P.S. Enjoying my content? ⚕️Follow me for more tips on medical device commercialization and repost to share with your network. #medtech #gotomarket #medicaldevices #healthcareinnovation

27

2 Comments

Revolutionizing Cancer Care with AI: Insights from Karyn Goodman Karyn Goodman, GI radiation oncologist at Icahn School of Medicine at Mount Sinai, and associate director of clinical research for Texas Cancer, highlights AI's role in radiation oncology. 📊✨ AI models are being developed to handle imaging and data tasks, such as contouring tissues on CT scans. This innovation enhances efficiency and precision in treatments, freeing up human expertise for more complex tasks. 🌟💡 Discover the transformative power of AI in cancer care! 🌐🔍 💡 Let's uplift each other and make wellness a priority: https://lnkd.in/gawjg45G #CancerTreatment #AIinMedicine #RadiationOncology #InnovativeHealthcare #TransformWellnessTogether #AIWellness #healthcare #healthandwellness

3

More news on the challenges of finding a business model for digital health companies

16

Like an unfazed young quarterback who keeps firing it downfield despite a spate of interceptions, CMMI keeps tossing around VBC models hoping something will connect. Contrary to its intent, CMMI's programs increased spending by $5.4 billion from 2011-2020 with another increase in net spending of $1.3 billion by 2030. Of the more than 50 models the Center has implemented, only 6 have generated "statistically significant" savings. Only 4 (Home Health Value-Based Purchasing, Prior Auth Model for Non-Emergent Ambulance Transport, Diabetes Prevention Program, and Pioneer ACO) were certified for expansion. CMMI points to the voluntary nature of the programs as a challenge to achieving cost savings. However, CMS hasn't done itself any favors here by aggressively ratcheting down payments and making downside risk >>> upside risk. (Notably, two upcoming programs -- TEAM and Inpatient THA/TKA PROMs are mandatory). Finding the VBC sweet spot was always going to require some trial and error. Perhaps CMMI has learned from the programs that failed or weren't chosen for expansion. TEAM doesn't seem to be all that different from CJR. Making Care Primary aims to closely integrate primary and specialty care (or, more likely, incentivize PCPs to take on more of this work). Not surprisingly, one of the criticisms is lack of provider input. The programs are too complex, too expensive to administer, and aren't transparent. CMMI should take note of several studies demonstrating the success of physician-led VBC efforts. Here are some other thoughts: - Offer financial grants in 5- or 10-year terms for any organization that designs its own VBC program and presents a reasonable and viable clinical/business model. CMMI would do itself a favor by partnering with or at least listening to innovators. - Medicare Advantage complexity and risk adjustment handwaving shouldn't be the only mechanisms by which advanced providers can profitably care for beneficiaries. That program isn't saving money either, but it's the only way many of these approaches can generate meaningful revenue (and most still aren't profitable). - Have regular meetings with those building VBC models, particularly those with strong physician leadership. Those meetings should be transparent and productive of action plans and iterative changes that increase model viability. - Be more open about what your goals and targets are. No one should be a victim of their own success, and if you keep moving the goalposts, you shouldn't be surprised when people get frustrated and drop out. I'm certain that, given fair incentives, a meaningful voice, and an equal seat at the table, healthcare innovators can design sustainable VBC programs. But there needs to be more collaboration and a willingness to listen and engage. Otherwise, you're just slowly adding to the alphabet soup of also-ran programs. #medicine #healthcare #health #vbc #healthcareinnovation #valuebasedcare

50

17 Comments

🚀 Digital Health Funding Soars by 48% in Q1'24! In Q1'24, digital health saw an impressive 48% increase in funding compared to the previous quarter. This surge outpaces the broader venture capital landscape, which recorded an 11% growth. While digital health was on the rise, other industries experienced more moderate gains or even declines: - Fintech: -16% QoQ - Retail Tech: +17% QoQ This uptick in digital health funding highlights the growing interest in technology-driven healthcare solutions. It's a sign that investors are recognising the potential for innovation and positive impact within this sector, even as other industries face more mixed outcomes. 🌱💰 #DigitalHealth #VentureCapital #StartupFunding #Innovation

4

Why Physicians Should Consider Investing in HealthTech Venture Funds As a physician deeply entrenched in the world of healthcare, I've observed firsthand the transformative impact of technology on our industry. From telemedicine to AI-driven diagnostics, the pace of innovation is staggering—and it’s only accelerating. Today, I want to share why we, as physicians, should consider investing in HealthTech venture funds 👇 - Driving Innovation: By investing in HealthTech funds, we're not just placing capital; we're actively propelling the future of medicine. Our investment helps fund the technologies that could one day revolutionize patient care and our daily medical practices. - Insider Expertise: Who better to understand the potential of a new medical technology than those of us in the field? Our clinical experience enables us to assess the feasibility and potential impact of innovations, making us well-suited to make informed investment decisions. - Spreading Impact Beyond Practice: While we impact lives daily through direct patient care, investing in a HealthTech fund allows us to extend our influence by supporting solutions that reach a global patient base, potentially benefiting millions. - Financial Returns: HealthTech is one of the fastest-growing sectors, and investing in this area offers a valuable opportunity for substantial financial returns. Diversifying our portfolios beyond traditional stocks and real estate into venture funds can offer a unique balance of risk and reward. - Personal and Professional Growth: Engaging with the startup and tech ecosystem exposes us to new ideas and business strategies, broadening our horizons beyond clinical practice. This can lead to professional growth and networking opportunities that may not otherwise arise. I encourage my fellow healthcare professionals to look into HealthTech venture funds as a way to participate in the broader spectrum of healthcare innovation. Our expertise and capital can play a crucial role in shaping the future of the industry. ---- To join our vibrant community of 2,000+ healthtech experts and thought leaders, click the link in my profile bio.

19

AI in healthcare is set to significantly enhance efficiency, improve patient care, and streamline operations. It's already showing promise in areas like radiology, where it has enabled a 50% efficiency gain in mammography. However, challenges such as re-certification of AI algorithms and integration into clinical workflows remain. The sector is witnessing growth, with FDA approvals increasing and mergers and acquisitions indicating a maturing yet still nascent market. #unitedseniorassociation #USA #eldercare #FDA #seniorliving #ADA #community #patientcare #compliance #hospitals #OIG #healthcare #digitalhealth #AARP #assistedliving #HSS #CMS #nursinghomes #seniorcare #medicare #seniorcitizen #homehealthcare #caregiver #dementia #inhomecare #seniorhealth #HHSOIG #kloekorby #healthcarefraud #healthcarepolicy #HIPPA #alzheimers #mentalhealth

5

1 Comment

One key reason why healthcare technologies like AI, telemedicine, and software haven't structurally lowered costs or broadened access for the entire system after decades of experimentation (excluding pure financial engineering of payer contracts) is the limited reimagining of care delivery itself, with asynchronous telemedicine as a notable exception. In the same way that digitizing physical paper failed to exploit the capabilities of a new computing medium for word processing, hosting physician-patient conversations over video does not transform the actual delivery of care. In telemedicine, many of the inputs (e.g., clinician time) and therefore cost structures of in-person care persist, which is why we often see fee-for-service telemedicine companies struggle to achieve economies of scale (i.e., growing revenues non-linearly while growing costs linearly). Until we deconstruct the components of a medical appointment – from intake and triage, to chart review, history gathering, physical examination, diagnosis, treatment, and follow-up – we will continue to be underwhelmed by the disconnect between the transformative potential of these technologies and real-world patient/system impact. I’m optimistic that the emergence of reliable and performant medical AI agents will urge us to perform this deconstruction and redesign the fundamental building blocks of a “medical appointment.”

215

39 Comments

Had an insightful conversation with David Rhew, M.D. and Missy Krasner on topics at the intersection of healthcare and AI. We discussed the right healthcare use-cases to go after in 2024, where to start, what not to do, and what governance looks like for Generative AI. Check out the video here: https://lnkd.in/g4KQ4kK5 #healthcare #AI #generativeAI #ai #digitalhealth #governance #llmgovernance

27

lol How is AI weaving its way deeper in to healthcare….🤔 As we navigate the evolving landscape of healthcare, the conversation around AI often gravitates towards its potential risks and downsides. However, there is also a different narrative—one where AI serves as a transformative tool for specialists making critical decisions about their patients’ and clients’ care. While some may fear that AI aims to replace healthcare professionals, the truth is far from that. AI is not about replacing the invaluable human touch that specialists and prodessionals provide, nor is it merely about speeding up processes. Instead, it’s about augmenting their capabilities, offering new perspectives, and providing a deeper understanding of each individual’s unique symptoms or scenario. AI tools like the ones we use at FirstHx empower regulated healthcare professionals and physicians with advanced insights and clarity. They offer an array of perspectives on how symptoms can be indicators or catalysts for various conditions, ultimately helping guide the professionals to make decisions around next steps in treatment or advice. This technology enables more focused and individualized care, ensuring that every action is tailored to the specific needs of the patient. By leveraging AI, healthcare professionals gain new tools that enhance their ability to serve their populations more effectively. It’s not about replacement; it’s about empowerment and improvement. AI in healthcare is about giving specialists the vision and clarity to make more informed, precise decisions, leading to better outcomes for all. Check out how we are assisting professionals strive for optimal outcomes at First HX. Here is our founder Mark Benaroia, MD, MHSc, FRCPC breaking down how the company and technology came to fruition in his latest appearance on the Maker’s Bar Podcast. https://lnkd.in/grna7gqf #AIinHealthcare #HealthTech #Innovation #PatientCare #HealthcareProfessionals #FirstHx #FutureOfHealthcare #mentalhealth #behaviouralhealth #AI #physician #psychology #BCBA #socialwork #medical

7

4 Comments

How long will it be until it’s malpractice to not be leveraging #GenAI in medical triage in the ER, in radiology & beyond? A study by Christopher Williams, Atul Butte et al from University of California, San Francisco and published in JAMA Network Open found that real world #AI #GPT accurately identified the patient w/ higher acuity when given pairs of de-identified presenting histories. The model could accurately prioritize the sickest patients 89% of the time, comparable to physicians who were correct 86% of the time. #LLM-assisted triage systems could help free up clinicians to treat the most serious patients first. JAMA Study: https://lnkd.in/gh-c5VYs https://lnkd.in/g8NUHPFm #triage #patientsafety #healthcare #generativeAI

103

20 Comments

Med Tech in Focus:   Piggybacking on Blood-Glucose Monitors for Other Types of Drugs Researchers at Rice University have modified blood-glucose monitors to make them applicable to a variety of drugs, including chemotherapy agents.  Current blood-glucose systems monitor specific proteins that bind to glucose molecules and give off electrons in a biochemical reaction. Those electrons then emit a current that indicates the level of glucose in the blood. The engineers at Rice used that same process to monitor levels of an anti-cancer drug, afimoxifine, an estrogen inhibitor that is also made after a patient takes tamoxifen. Testing more than 400 modified versions of proteins revealed a version that reacted with afimoxifene, releasing electrons, and producing a current in the same way that glucose systems did. The development shows promise for monitoring and dosing other types of drugs and biomarkers in the blood and moves away from one-size-fits-all towards a more personalized approach to drug dosage. Because the methodology is the same, a patient can still test with a strip or a skin monitor and store or share data in real time. Further, the new technology piggybacks on devices that are inexpensive, readily available, and familiar to consumers.  Investors are always adjusting their "Play Book" with a variety of options for investing. Mikal Ventures is one of those alternatives, offering curated solutions to real world problems. Reach out if you have questions on how we can be of assistance and bookmark https://mikalventures.com/ for additional insights and information.

2

1 Comment