Healthcare Innovation Models

New research prototype for Personal Health Agent (PHA), a comprehensive research framework for delivering personalized, evidence-based health and wellness guidance. This system is built on a multi-agent framework that models support after a human expert team, each handled by a specialized LLM sub-agent: ▶️ Data Science Agent: Analyzes multi-modal data from wearables and health records, such as blood biomarkers, to provide contextualized numerical insights. ▶️ Domain Expert Agent: Acts as a reliable source of grounded health knowledge, tailoring information based on the user's specific health profile. ▶️ Health Coach Agent: Supports users in goal-setting and behavioral change through multi-turn, psychologically-inspired conversations. The Orchestrator dynamically coordinates these specialists to synthesize a single, coherent response to complex queries. Evaluations confirmed that this collaborative multi-agent approach significantly outperformed single-agent baselines in overall response quality, clinical significance, effectiveness and usefulness as evaluated by human experts and end-users. This work, including extensive evaluation of all agentic components using the Wearables for Metabolic Health (WEAR-ME) study data, establishes a validated blueprint for the next generation of trustworthy and coherent personal health AI. Read more about this research and the multi-agent framework: https://goo.gle/42kzjvZ Preprint: https://lnkd.in/dfZ96X5c

Turns out the best investment opportunities in healthcare are hiding in nausea, gallstones, and constipation. In recent conversations with several large healthcare private equity funds, the consensus is clear: GLP-1 drugs like Ozempic, Wegovy, and Mounjaro are not just transforming obesity care. They are quietly creating a secondary gold rush in treating the side effects of weight loss. And it is not small. More than 6 million Americans are already on GLP-1s. Global GLP-1 sales are expected to exceed 130 billion dollars annually by 2030. Up to 40 percent of patients experience significant gastrointestinal side effects. Around 5 to 7 percent end up with gallbladder complications. Sarcopenia is now a real concern. Mental health utilization among GLP-1 users is rising by nearly 20 percent. Fertility clinics are seeing double-digit growth from GLP-1-related cases. So while everyone is applauding the miracle of weight loss, the savviest investors are looking at the flip side. They are rolling up GI clinics, expanding ASC platforms with cholecystectomy capacity, funding digital fitness and nutrition programs to fight muscle loss, backing behavioral health services for body image and binge relapse, and building analytics tools to help payers track it all. It is not just a new drug. It is a new healthcare economy. I have unpacked this trend in detail in my latest white paper: a playbook for investing around the GLP-1 explosion by targeting the ripple effects no one is talking about. Because in healthcare, what goes down (appetite) must come up (utilization of something else). #PrivateEquity #Healthcare #GLP1 #OzempicEconomy #HealthTech #DigitalHealth #VentureCapital #ObesityCare #PEInvesting

HealthTech AI is no longer exciting. It’s expensive. And the market has re-priced itself for performance. The first half of 2025 solidified a new reality in digital health. US-based digital health startups secured $6.4 billion across 245 deals (Rock Health). While total funding is up from H1 2024, the trend of fewer, larger checks persists. Rock Health pegs the average deal size at a robust $26.1 million, a significant increase from $20.4 million in 2024, signaling a concentrated investment in more mature, impactful companies. Investors are no longer buying potential. They're buying precision and demonstrable value. They care if your AI: Saves hours, not just clicks: The focus is on quantifiable time savings for clinicians and administrative staff, directly addressing burnout and efficiency gaps. Cuts costs, not just code: Real-world cost reduction is paramount, whether through optimized operations, reduced errors, or improved resource allocation. Embeds in real workflows, not pitch decks: Solutions need to be seamlessly integrated into existing healthcare systems, proving their utility in daily practice. McKinsey calls this the "productivity premium," and it has become the new funding filter. A significant portion of VC dollars continues to flow into AI-enabled startups, not because they're novel, but because they perform and deliver tangible returns. Abridge: This AI note-taking startup for doctors raised a staggering $316 million in June 2025 (Series E), bringing its total funding to over $770 million. Its value proposition is clear: giving clinicians hours back by automating documentation. Innovaccer: Secured $275 million in Series F funding in January 2025 to expand its AI and cloud capabilities, aiming to be a "one-stop shop" for healthcare AI solutions. They focus on data aggregation and intelligence to optimize value-based care programs and reduce administrative burden. Truveta: Raised $320 million in Series C funding in January 2025, solidifying its position in health data and analytics. Their mission revolves around leveraging data to drive insights and improve care. Hippocratic AI: Completed a $141 million Series B financing round in February 2025, valuing the company at $1.64 billion. Their focus is on developing safe, patient-facing AI for non-diagnostic tasks, addressing healthcare staffing shortages. These companies optimize operations, not optics. The delta? Execution. This is not a hype cycle. It’s a competency correction. The end of vision-only founders. The rise of operator-founders who understand: Unit economics: The true cost and value generated by each patient interaction or service delivered. Integration latency: The speed and ease with which new technologies can be embedded into complex, often legacy, healthcare IT infrastructure. Reimbursement drag: Navigating the intricate and often slow process of getting innovative solutions covered by payers. What part of this feels uncomfortably true?

Let’s talk about a revolutionary approach to healthcare. Systems science is reshaping how we view health by highlighting the interconnectedness of the biological, psychological, and social factors that influence well-being. In holistic medicine, this framework is more relevant than ever. Here’s why it works: → Holistic Perspective Health isn’t just physical. By embracing the interconnected nature of the body, mind, and environment, we can understand how each part affects the whole person, leading to better care outcomes. → Complex Adaptive Systems Health is complex and dynamic. Genes, lifestyle, and environment interact in non-linear ways—systems science helps us understand how these factors shape overall health, making personalized care more effective. → Dynamic Interactions Health isn’t static. It’s shaped by constant interactions between internal and external factors. By understanding these dynamics, we can prevent diseases before they manifest and provide more tailored treatments. So, how does this impact patient care? → Personalized Plans Systems science enables healthcare providers to create individualized plans that address the root causes of health issues and predict future trajectories that can be optimized or prevented. → Interdisciplinary Collaboration Healthcare isn’t one-size-fits-all. By encouraging collaboration across various disciplines—medicine, psychology, nutrition, and public health—we can create comprehensive care plans for patients that lead to better results. → Preventive Healthcare A systems approach helps identify risks early, allowing for proactive strategies that can prevent diseases from developing. The takeaway? Systems science brings a holistic, interconnected perspective to medicine, enabling us to better understand, treat, and prevent health issues. Are you ready to embrace a more complete, systems-based approach to healthcare? Let’s dive in.

I’m just old enough to have practiced medicine in an analog world, where pen and paper naturally constrained information. There was only so much we could write. I remember how my hand would ache after writing an H&P and admission orders (especially if including an insulin sliding scale or heparin drip). Note bloat wasn’t an option. Information didn’t flow freely either. Learning what had happened elsewhere meant trekking to the medical records department or waiting by a fax machine. Often, it was easier to triangulate and infer the history. We know what happened next. EHRs have brought enormous benefits, but also information overload. The average medical record is now half the length of Hamlet. Our inboxes overflow with innumerable patient messages, results, and notifications. We need help. Fortunately, LLMs are very good at summarizing and transforming information. Tech analyst Scott Belsky forecasts an impending “Era of Summarized Living,” where nearly every interaction is condensed for later recall. We're already seeing this to an extent. After Zoom calls end, we now receive automated summaries with clear action items. Healthcare is heading the same way. AI scribes already summarize conversations into notes. We will soon prepare for visits by reading CliffsNotes-style digests rather than raw records. EHRs will provide contextual summaries alongside results and patient messages. This will help. And it will also change us. We’ll be one degree removed from the source information (which we would often not have reviewed anyway). We will write different kinds of notes (and may eventually stop writing them at all). We may even start speaking differently to optimize what the summaries capture (“you’re making an important point the summary should include”) . Ultimately, we may develop a kind of shared institutional memory that anyone can access. But there will be harder questions. Like what gets captured? What information gets lost? And how much of our thinking and reasoning is tied to today’s manual processes? This shift is a necessary response to the overload digitization created. I'm mostly excited about it. But I also wonder what happens when the summary becomes the reality, and whether we will maintain the discipline to return to the source when it matters.

🔬 Biostatistics & Epidemiology: A Powerful Partnership in Public Health 🧠📊 Epidemiology and biostatistics are not just allied disciplines—they’re inseparable pillars of modern public health. Their synergy fuels scientific discovery, guides policy, and shapes interventions that save lives. 🎯 Complementary Roles 🔍 Epidemiology: The Question Generator Epidemiologists identify public health problems, generate hypotheses, and design studies to explore disease patterns and risk factors. • Study Design: Choosing between cohort, case-control, or cross-sectional designs. • Hypothesis Generation: Spotting unusual disease clusters and exploring environmental or genetic links. 📈 Biostatistics: The Analytical Engine Biostatisticians bring rigor and structure to analysis, transforming raw data into actionable insights. • Data Analysis: Regression models, survival analysis, hypothesis testing. • Bias Control: Using stratification, matching, and advanced techniques to ensure validity. ⸻ 🔄 Collaboration Across the Research Cycle From idea to impact, these fields work hand-in-hand: • Design: Epidemiologists define the framework; biostatisticians handle power calculations and randomization. • Data Collection: Field implementation meets data integrity. • Analysis & Interpretation: Quantitative models meet contextual insight. ⸻ 🌍 Landmark Public Health Successes • Framingham Heart Study — Identified key cardiovascular risk factors. • Polio Vaccine Trials (1954) — Pioneered RCTs and statistical validation. • COVID-19 Response — Modeling and analytics shaped global strategies. ⸻ 🔬 Modern Integration & Innovation Today’s research integrates both fields in exciting ways: • Causal Inference — From associations to causation using tools like mediation analysis and instrumental variables. • Big Data & Machine Learning — Leveraging EMRs and AI for scalable insights. • Precision Medicine & Spatial Epidemiology — Mapping risk at individual and population levels. ⸻ 🚧 Challenges & 🚀 Opportunities Challenges: • Managing high-dimensional data. • Bridging training gaps between disciplines. Opportunities: • Interdisciplinary education. • Harnessing AI to merge epidemiological context with statistical power. ⸻ 🧠 At the intersection of inquiry and evidence lies the true strength of public health. As biostatistics and epidemiology evolve together, they continue to shape a healthier, data-driven world. 🌐 #PublicHealth #Biostatistics #Epidemiology #Research #DataScience #CausalInference #HealthPolicy #PrecisionMedicine #AIinHealthcare #LinkedInScience

🚀 Medtech VC Investment: A Mixed Bag or a Sign of Growth? In Q3 2024, global medtech VC investment dipped to $3.2 billion, down from $3.6 billion in Q2. 📉 But before we sound the alarms, here’s why it’s not as concerning as it seems: 👉 VC deal activity was up 9.4% YoY, showing resilience despite typical Q3 seasonality. 👉 2024 deal flow is 17% ahead of 2023, reinforcing the view that last year marked the low point in the funding cycle. Yes, quarter-to-quarter fluctuations are expected, but the broader picture remains one of recovery and growth for the medtech sector. #medtech #vc #exits #medicalinnovation 💼 Exit Outcomes Shine Bright in Q3 While funding showed some choppiness, exit activity stole the spotlight: High-profile acquisitions of Endotronix, Paragonix, Innovalve, Endomag, and EndoGastric Solutions – all valued at over $100M. 💰 ⚙️ M&A: The Exit Strategy of Choice With IPO markets still challenging, M&A continues to dominate as the preferred exit route. But there’s optimism on the horizon – 2025 could see an IPO revival, as the backlog of startups nearing public readiness builds. 💡 Key Takeaway Medtech is holding steady in a volatile environment, and while quarter-to-quarter shifts in funding are inevitable, the upward trajectory remains clear. 🚀 The sector’s adaptability, bolstered by strong exits, is paving the way for a promising 2025.

How can we decrease pharmacy spend on high-cost drugs by double digits without worse outcomes? --- Uplift modeling is a common tactic in marketing to target the specific people for a promotion that otherwise wouldn’t buy the product. While marketing in general can lead to overconsumption, in healthcare/#pharmacy, the same mathematical techniques used for uplift modeling could be repurposed to support #PrecisionMedicine or personalized medicine, where the goal is to identify which patients are most likely to benefit from a specific treatment while avoiding unnecessary treatments for patients who might not respond well. Identifying the cohort that is getting most of the outcomes from a drug varies by drug, but some drugs have only a fraction of the total population driving a larger share of clinical results. --- Here's the basic process for using #UpliftModeling (you can find more details from my Milliman white paper in the comments): 1. Treatment: Identify the treatment for which you want to predict response (e.g., a high-cost brand/specialty drug like GLP-1s). This could also be done for a medical device or any intervention. 2. Data collection: Gather comprehensive data and studies about patients, including their medical history, genetic information, and any other relevant attributes. This is often the limiter of building a good model. 3. Control group: Assemble a control group of patients who are similar to those receiving the treatment but are not receiving the treatment themselves. This helps establish a baseline for comparison. 4. Outcome measurement: Measure the effectiveness of the treatment for both the treatment group and the control group. This could involve monitoring health improvements, cardiac events, or other relevant medical outcomes. For FDA-approved drugs, this could come from published research on the “absolute risk reduction” or “number needed to treat.” 5. Model building: Develop predictive models using machine learning algorithms that estimate the likelihood of a positive response to the treatment for each individual. 6. Uplift calculation: Calculate the difference in response rates between the treatment group and the control group to determine the net impact of the treatment. 7. Segment: Divide patients into different segments based on their predicted response probabilities. 8. Action: Use the insights from uplift modeling to guide treatment, coverage, or other decisions. --- A payer or employer can use this information how they’d like, but I imagine it will be used to adjust formularies or utilization management strategies. It could also be used when setting up contracts for how a drug should be used while carving out certain drugs or disease states (e.g. oncology drugs at a center of excellence). There are more potential use cases in the white paper in the comments. --- Would you use this strategy for #PharmacyBenefits or #ValueBasedCare models that take on risk for cost of care?

Healthcare isn’t sexy. But in 2025, it’s quietly becoming the smartest play in Indian VC. India's VC funding grew 40% in Jan–Feb 2025, even as the global markets cooled. Within this surge, healthcare has emerged not just as a resilient sector but as one of the most strategically favoured verticals. This shift isn’t incidental. It’s being driven by a very real behavioural pivot. 1) Indian consumers, particularly in Tier 1 and urban Tier 2, are moving decisively from reactive to proactive healthcare. 2) There is growing willingness to pay for diagnostics, preventive care, and ongoing wellness services. Healthcare is no longer something to turn to in crisis — it's becoming a monthly subscription decision, a daily app notification, a data-driven lifestyle choice. On the delivery side, investors are moving away from capital-heavy multispecialty hospitals toward scalable, asset-light models. Single-specialty chains in categories like fertility, ophthalmology, orthopaedics, and dermatology are becoming favoured targets. They offer sharper operational visibility, lower complexity, and faster paths to profitability, all of which align perfectly with current VC risk appetite. Diagnostics, too, is undergoing a transformation with significant interest in digital-first, at-home, and B2B diagnostics platforms. When combined with health SaaS, workflow automation, and AI-enabled diagnostics, the thesis becomes not just consumer-facing health, but infra-first healthcare. Parallelly, India’s medtech and pharmaceutical manufacturing sectors are benefiting from global realignment. As global players diversify away from China, India’s high-quality, low-cost base is becoming strategically important. Private equity interest is rising in CDMO platforms, API manufacturers, and device exporters. These businesses offer forex-linked revenue, regulatory clarity, and strong M&A appetite from strategic buyers. Across the board, the VC lens in 2025 is more thesis-led and less trend-driven. Investors are optimising for clear profitability paths, clean cap tables, recurring revenue models, and compliance hygiene. Healthtech models with real operating leverage are being favoured over B2C plays chasing vanity growth. In short, this is not a post-pandemic sector play. It’s a structural investment shift. Healthcare in India is maturing, in demand, delivery, and investor mindset. #100DaysLinkedIn #healthcare #VC

Academic research published during my 10+ years at Duke Institute for Health Innovation was cited over 1,000 times in 2025. Some unspoken lessons about interdisciplinary research: 1) rejection is the norm - My first first-author paper was rejected 7 times before getting accepted (https://lnkd.in/eDENrp9M). Across all publications, I would guess each publication is rejected on average 3+ times. I've only ever had one publication accepted without revisions (https://lnkd.in/eYyQfirF). 2) credit is infinitely divisible - Eagerly collaborate and include students, trainees, clinicians, and administrators in your work. Value and elevate the work that takes place to implement innovations. Two notable examples from our own work: publishing the implementation of the Sepsis Watch model with 27 collaborators (https://lnkd.in/eBXKPyGq); publishing a review on eliminating the digital divide with 50 collaborators from Health AI Partnership (https://lnkd.in/e5Bjkbiv) 3) learn the publishing norms from different disciplines - Stats / CS publish full-length manuscripts at conferences. Law review and social science articles are typically one, two, or three authors. Challenge publishing norms where you can and respect norms when you must. Be willing to provide extensive feedback on written outputs to collaborators in different disciplines, even if you can't be a named co-author. 4) peer review is mostly random - If we get critical reviews and a rejection, I typically resubmit the paper as-is to a new publication outlet. The chances two different sets of peer reviewers have the same feedback is exceedingly low. Focus effort on revisions when you have a path to publication with the same reviewers. 5) submit to new journals - MLHC, FAccT, Nature Digital Medicine, and PLOS Digital Health were all started in the last ~15 years. Don't be afraid to support new journals that cut across disciplines, especially if the journal aligns with your interdisciplinary research and interests. That journal could be big some day! 6) leverage invitations to publish - If you are invited to write a piece or edit a collection, leverage the opportunity! In 2020 we wrote an invited review that's now been cited 200+ times (https://lnkd.in/eqv_B7jt) and in 2023 we curated a collection of manuscripts that's now been viewed ~70,000 times (https://lnkd.in/e2EHdB8a). 7) invest time building skills - I spent 4 years working on healthcare data science / ML projects before participating in my first publication (https://lnkd.in/e-pENwA8). Productive != Publications. Productive can mean building skills that enable you to conduct groundbreaking research in the future.