Jassi Pannu

Check out a new JAMA Viewpoint article on state biomarker testing laws that I co-authored with UCSF colleagues Grace Lin, MD, MAS, and Kathryn Phillips, PhD, of UCSF's Center for Translational and Policy Research on Precision Medicine (TRANSPERS). I've reposted TRANSPERS news release below. MON MAY 13, 2024 In the JAMA article, The State of State Biomarker Testing Insurance Coverage Laws, by Lin et. al., the authors tackle the complicated topic of states legislating access to medical care. In response to perceived barriers to access to care, states have enacted legislation to mandate terms and conditions of insurance coverage for a variety of circumstances, such as cancer screening and mental health treatment parity. Recently there have been increasing numbers of states that have passed legislation mandating that insurers cover “biomarker testing”. This review of legislation from 15 states mandating coverage of biomarker testing finds that their impact may be limited by their lack of reach and implementation challenges. While these laws attempt to expand access to coverage, that may not always be the case, since more than 50% of the population may not be covered by legislative mandates governing insurance coverage. Additionally, such insurance mandates may not affect inequities in access. Finally, there is little guidance for implementation in the legislation. The authors conclude that current laws are likely to have limited effect, particularly concerning improving health equity, and additional approaches will likely be needed to improve patient access and outcomes. This study was led by Grace Lin, MD, MAS (UCSF School of Medicine) and TRANSPERS (UCSF Center for Translational and Policy Research) founding director Kathryn A. Phillips, PhD, with Janet Coffman, PhD, MPP, MA (UCSF Institute for Health Policy Studies). TRANSPERS is in the Department of Clinical Pharmacy at UCSF and is affiliated with the Philip R. Lee Institute for Health Policy Studies.

23

The rise in lung cancer rates among female non-smokers might be linked to radon inhalation. This theory is based on the differences in radon exposure between males and females, which can be attributed to the variations in time spent indoors versus outdoors.  Source: Safari, A., et al. (2024). "Exploring the Disparity in Indoor/Outdoor Time and Radon Exposure as Possible Factors Contributing to the Unexpected Increase in Lung Cancer Risk among Non-Smoking Women." Journal of Biomedical Physics and Engineering. in press. https://lnkd.in/ewvumYy9 #lung #cancer #females #nonsmokers #radon #Indoor #Outdoor

16

1 Comment

Celebrating Dr. David R. Liu's Pioneering Work in Gene Editing at PMWC 2025! We are excited to honor Dr. David Liu at PMWC 2025 (February 5-7, Santa Clara Convention Center). His groundbreaking work in base editing and prime editing has revolutionized genetic medicine, offering new hope to millions. An Exclusive Q&A with Dr. David Liu Q1: Can you explain the difference between base editing and prime editing? Dr. Liu: “While both base editors and prime editors enable precise changes to DNA, base editing directly converts one base into another (for example, A to G) without causing double-stranded breaks. In contrast, prime editing can perform a broader range of edits, including insertions and deletions, offering more versatility in precise DNA modifications.” Q2: How is your work transforming clinical applications? Dr. Liu: “The power of base editing was highlighted when it was used to treat T-cell leukemia in a young girl who had no other options. We used base-edited CAR T-cells, and her leukemia went into complete remission—a powerful testament to the real-world potential of these technologies.” Why Dr. Liu's Work Matters: -Base Editing enables precise changes to DNA, making it possible to correct mutations that cause genetic diseases like sickle cell anemia and cancer. -Prime Editing extends these possibilities even further, allowing scientists to insert or delete sections of DNA to treat more complex genetic conditions. Dr. Liu’s work is at the forefront of medical innovation, offering a future where precise gene correction is a reality for millions of people suffering from genetic diseases. Join me in congratulating Dr. Liu on this well-deserved recognition as the PMWC 2025 Honoree! 👏 Beam Therapeutics, Amy Simon, Prime Medicine, Inc., Editas Medicine Exo Therapeutics, Broad Institute of MIT and Harvard, Gaddy Getz, Anahita Vieira, Ph.D., Emily Botelho, Michael Bruce, Gilmore ONeill, James Mullen, Alix Ventures, Nicole M. Gaudelli, Luke Koblan, Janice Chen, Mammoth Biosciences, University of California, Berkeley, Fyodor Urnov, Jacob Herman, Ph.D., Fadi Najm, Jonathan Gootenberg, Charles Gersbach, Prashant Mali, Nadav Ahituv, Morten Sogaard, Trevor Martin, Shilpi Arora  Dr. Liu will be giving his keynote on the future of gene editing technologies and their growing clinical applications. Don’t miss the chance to hear from one of the brightest minds in the field! See program: https://lnkd.in/gm9nsW_G PMWC - Precision Medicine World Conference,

79

Important data from The Commonwealth Fund reports "Mirror, Mirror: A Portrait of the Failing US Health System" (https://lnkd.in/eAgupmSX) in which they conduct a comprehensive analysis of 70 health systems in 10 countries. The US is dead last in overall care, access to care, and health outcomes and nearly last in equity and administrative efficiency. One of their key conclusion points about our failing system? LACK OF INVESTMENT IN PRIMARY CARE. "Years of neglect and under-compensation for primary care have resulted, predictably, in nationwide shortages of the clinicians who play a vital role in managing chronic illness and reducing the need for costly and sometimes unnecessary emergency, specialty, and hospital care services. The acquisition of primary care practices by health systems and private equity investors is further disrupting an already fragile primary care capacity, with uncertain short- and long-term consequences. The fragmented nature of the U.S. health care system makes it difficult even for many well-insured patients to access convenient and effective care." These results square with everything that patients, my colleagues and I know. How much more data do we need to transform primary care in our country?

31

2 Comments

The important milestone of medical AI development was highlighted by Eric Topol, MD in his Ground Truth publication "When Medical A.I. is Lifesaving. Counterbalanced by some disappointments". At the end of April, research group from Taiwan published a paper describing results of a  multisite randomized controlled trial evaluating the ability of an AI-enabled electrocardiogram analysis tool to identify hospitalized patients with a high risk of mortality. The study was appeared in Nature Medicine and included 39 physicians and 15,965 patients. Electrocardiograms of patients in interventional arm were analyzed with AI and physicians received alert with report and warning messages, flagging patients predicted to be at high risk of mortality. The control arm included patients whose ECGs were analyzed by conventional method (by a physician with no AI involvement). The implementation of the AI-ECG alert was associated with a significant reduction (17%) of all-cause mortality within 90 days and saving 7 lives per 100 in some groups. As Dr. Topol pointed out, the results are comparable to most successful medical treatments, such as statins. Notably, it was the first-ever RCT to estimate the effect of AI-based intervention on patients' mortality. However, the results of four trials of real world scenarios of Generative AI use, as described in the same Ground Truth publication, were not so promising.  Large Language Models (including Chat GPT 4) were unsuccessful in answering oncology questions and assisting in medical coding. They also failed to reduce clinicians’ time spent writing reply messages to patients (though they did generate more empathetic replies) and posed a risk of severe harm in 7% of generated messages which is definitely not acceptable. Surely, as Dr. Topol stated “Progress in medical A.I. won’t occur in a straight line”. It has been repeatedly confirmed that AI is far better than humans at analyzing repeated patterns not generated by humans, such as MRI or ECG. However, when it comes to working with more abstract entities, such as diagnoses or clinical messages, its effectiveness is much more modest. In any case, I strongly encourage you to refer to the original Ground Truth publication for more details - https://lnkd.in/dXaEAq6m

21

2 Comments

I am excited to share this paper co-authored by 20 PROMIS National Centers across the world. We present a harmonized global approach to international standardization of PROs and PROMs, based on the #PROMIS measurement system. There is growing evidence that outcomes such as pain, fatigue, anxiety, depression, sleep disturbance, physical function, and the ability to participate in social roles and activities are relevant for most people, irrespective of their health condition. Measuring these outcomes routinely in all patients with PROMIS could increase outcome comparability and utility for a range of stakeholders. https://lnkd.in/eTdQJ89T

159

6 Comments

How will AI change the relationships between patients, caregivers, and doctors as each is empowered with more information and understanding? Patients, caregivers, and doctors will come into their visits a lot more prepared. For patients and caregivers: AI-enabled software agents will be used to support you in seeking better outcomes through personalized care. Healthcare often involves making complex medical decisions within a challenging environment: a balkanized medical system of many specialists, information overload and obsolescence, limited time with doctors, and siloed data. New AI tools can enable you to more actively participate in your treatment decisions. For example, consider receiving a complex scan report in a patient portal. Today you may be confused trying to interpret it, adding to your anxiety (“scanxiety”). However, a personalized generative AI agent can translate the scan report into language you can understand, and even contextualize it within your personal health history and clinical evidence/guidelines. By providing an understandable version of the report and the clinical context of your test results, you can engage more fully in decision-making with your medical team. For providers: - Visit preparation: The physician will be armed with the latest data and information, specific to that patient, e.g., test recommendations, treatment recommendations, referral recommendations, drug-drug interactions. - Notifications: Continuous notifications will come from patients and caregivers, e.g., from monitoring devices, and the medical team will have to respond to them. AI can triage the notifications and route them to different members of the medical team. - Treatment access: AI will help doctors in writing the letters that enable patients to access non-standard treatments. For more, please see our conversation with Ezra Cohen, MD, the Chief Medical Officer of Oncology at Tempus AI here: https://lnkd.in/erUhh4b7

24

2 Comments

Intellectually invigorated after attending Stanford University School of Medicine's first symposium on the role of AI in bio-medicine, convened by Dean Lloyd Minor and Fei-Fei Li (Co-director, Stanford Human-Centered AI Institute). Learned from leaders at Google, Amazon, Microsoft, American Medical Association, National Academy of Medicine, The National Institutes of Health, NHS, and eminent data scientists and ethicists at Stanford University. My Five KEY take-aways on successfully integrating AI systems into Healthcare. 1. TRANSPARENCY - Clinicians should know WHEN AI integrates into the systems they use. Failure to do so can lead to unintended consequences and patient harm. Example: Boeing made a change in the software in their new automated system, the Maneuvering Characteristics Augmentation System (MCAS), introduced with the 737 Max model. Pilots were either unaware of this change or not adequately trained on it, leading to fatal crashes and loss of life. 2. FUNCTIONALITY - Does it work? How well does it do what it is supposed to do? Post-use validation is paramount to refining systems. Academic and scientific rigor is important during building systems. However, it is unfair to ask our AI systems to be perfect prior to clinical implementation. Instead, the driving question should be - Does this particular AI system improve upon our current standard of care? 3. EXPLAINABILITY - Clinicians should know HOW AI integrates into their systems, and should be involved early on in the designing of these AI modules. End-user clinicians should know the LIMITATIONS and BIASES inherent to various AI modules, so they can deploy them knowledgably. Example: The dermatologist is aware that an AI trained on skin lesions of Caucasian patients may not have hi-fidelity when diagnosing lesions of dark-skinned individuals. 4. SCALABILITY - Health AI module development requires a lot of capital investment, often in the tens of millions of dollars. Most health AI modules are developed based on single institution specific data, and may not be applicable generally. Pooling credible data from multiple institutions/hospitals and using software to scale it for general use will bring down costs and increase adoption. 5. LEGAL/REGULATORY - Who is responsible when an adverse event happens as a result of implementation of an Health AI module- clinician, company, regulatory body? What regulations need to be put in place to deter nefarious actors from taking advantage of AI health modules? BONUS: ENVIRONMENTAL FRIENDLY? - AI systems are extremely water and energy intensive. How do we balance the benefits of Health AI with the adverse environmental impact the deployment of these models may have? AI in Healthcare AI in Healthcare Initiative (AIHI) Stanford Neurosurgery AI in Healthcare

38

We are thrilled to share our latest work titled "Mitigating the Risk of Health Inequity Exacerbated by Large Language Models". In this study, we find that the most widely used Large Language Models (LLMs) like GPT-4 and others may exacerbate health inequities, particularly in medical applications such as clinical trial matching and medical question answering. Our study was driven by the hypothesis that LLMs, trained on internet-based text data, inherit the disparities and inequities hidden and present in human society and history. Our findings confirmed this hypothesis. We demonstrated that when non-decisive socio-demographic factors, such as race, sex, income, LGBT+ status, homelessness, illiteracy, disability, and unemployment, are included in the input to LLMs, the models can produce incorrect and harmful outputs for these groups. If LLMs are widely adopted in healthcare, these discrepancies risk amplifying existing health disparities. To address this critical issue, we introduce EquityGuard, a novel framework leveraging contrastive learning to systematically evaluate and mitigate these risks. The framework ensures equitable outcomes across diverse populations, enhancing the health equity in healthcare applications of LLMs. Kudos to PittNAIL lab students' teamwork Yuelyu Ji, Wenhe Ma, Sonish Sivarajkumar, Hang Zhang, Xizhi Wu Check out the full paper here: https://lnkd.in/etinsdJ3 #HealthEquity #AIinHealthcare #LLMs #ClinicalTrials #GenerativeAI #ArtificialIntelligence

67

1 Comment

Read full paper in #Med

This landmark publication by the FDA's OC in JAMA illuminates the way forward in the oversight of clinical Generative AI (GenAI) solutions. The FDA leaders underscore the potential risks associated with these solutions while acknowledging their statutory limitations in regulating a technology that is profoundly affected by the context in which it operates. Check key excerpts from the FDA article: " …as that (Pre-Cert) program demonstrated, successfully developing and implementing such (Total Product Life Cycle) pathways may require the FDA to be granted new statutory authorities. The sheer volume of these changes and their impact also suggest the need for industry and other external stakeholders to ramp up assessment and quality management of AI across the larger ecosystem beyond the remit of the FDA." "The complexity of LLMs and the permutations of outputs necessitate oversight from individuals and institutions in addition to regulatory authorities. Because we cannot unduly burden individual clinicians with such oversight, there is a need for specialized tools that enable better assessment of LLMs in the contexts and settings in which they will be used." "Given the capacity for 'unlocked' models to evolve and AI’s sensitivity to contextual changes, it is becoming increasingly evident that AI performance should be monitored in the environment in which it is being used. ... The tools and circumstances of this ongoing evaluation must be recurrent and as close to continuous as possible, and the evaluation should be in the clinical environment in which it is being used." "Currently, however, neither the development community nor the clinical community is fully equipped for the recurrent, local assessment of AI throughout its life cycle. Health systems could fill this role, but currently their clinical information systems are unable to monitor the ongoing and long-term safety and effectiveness of these interventions." We agree that the way forward for overseeing clinical GenAI solutions will require a shift from a purely centralized regulatory model to a hybrid one that incorporates local, context-specific monitoring with a ground-level risk mitigation approach. Health systems can be equipped and incentivized to become widely distributed GenAI safety, effectiveness, and alignment sentinels using ISO 42001-certified AI Management Systems (AIMS). The pre-print articles linked in the comments detail how mutual assurance based on certified AIMS can become a foundation for clinical GenAI oversight. This is the adaptive approach we need to safely harness GenAI's potential in healthcare at scale. FDA Troy Tazbaz Robert Califf Haider Warraich ISO/IEC Artificial Intelligence (AI) Assistant Secretary for Technology Policy ANAB - ANSI National Accreditation Board U.S. Department of Veterans Affairs Coalition for Health AI (CHAI) Health AI Partnership National Institute of Standards and Technology (NIST) #GenAI #AIRegulation #AIMS #ISO42001 #AISafety

53

8 Comments

🌟 Thanks Kevin M. Baumlin, MD and the University City Science Center for inviting me to speak with your Fabulous Founders Fellows about building AI solutions in healthcare! Here are some key takeaways for aspiring founders: 🩺 Build With Clinicians, Not Just For Them: Collaboration with healthcare professionals ensures practicality, relevance, and operational readiness. 📈 Clinicians Demand Outcomes, Fewer Clicks, and ROI: Solutions must be efficient and demonstrate clear benefits. 🌊 Know Your Environment: Even the best ship will face a shipwreck if it doesn't understand the waters it's navigating. 🤝 Develop Deep Empathy for Stakeholders: Understand the needs and pain points of all involved. 🚀 Solve a Problem, Don’t Just Build ‘Sexy Tech’: Focus on creating meaningful solutions rather than just innovative technology. 🎯 "Value is Quality Over Cost: Know how to define and deliver quality in healthcare solutions," Kevin M. Baumlin, MD What are other words of wisdom from physician innovators?! Patrick B. Thomas, MD, MMCi, FACS, Hansa Bhargava MD, Harvey Castro, MD, MBA., Alfred Atanda Jr, Shrujal Baxi, Julian Mitton, MD MPH, Dr. Marc Gruner, Alonzo Sexton, Jennifer Miles-Thomas, Alyssa Abo, MD, MBA #AIinHealthcare #Innovation #HealthcareTech #Startups #HealthcareInnovation https://lnkd.in/esgfrmsr Jun Beom Pyo, Chinaemerem Daniel, Cole Hagen, JM Kim, Afsaneh Razi

33

6 Comments

New Paper ((https://lnkd.in/ew9eRZun, Stephen Gilbert Jakob Nikolas Kather Aidan Hogan) ) on a hot theme 🔥 : "Augmented non-hallucinating large language models as medical information curators". Large Language Models (LLMs) are opinion polarisers - their potential application creates arguments within and between physicians, developers, researchers, regulators and patients: - EITHER of extreme value and our entire future / OR of no value and a distraction; - EITHER can improve medical diagnosis and patient safety / OR a safety disaster waiting to happen; - EITHER a step change in medical efficiency / OR of another potential stressor for doctors and waster of their time; - EITHER a next big thing in medical technology entrepreneurship / OR a hype and a bubble about to burst. This paper wades into this debate, all guns blazing … and maybe its arguments will slightly annoy both the pro-LLMers and contra/anti-LLMers. - are you medical pro-LLMer? maybe this paper has a controversial argument for you as it explores mixing „old“ „epoch-1" AI technologies (Knowledge Graphs, as a form of Retrieval Augmented Generation - RAG) with your favourite shiny new LLM approaches, as this might just satisfy stringent regulations, that might demand all of determinism, lack of hallucination and a model of truth (see https://rdcu.be/dfLOJ) - are you and a contra/anti-LLMer? maybe this paper has a controversial argument for you as it allows that (augmented) LLMs might have a (very) central future in regulatory approved medical devices Exactly the question addressed in this paper, i.e., "does Retrieval Augmented Generation of LLMs through Knowledge Graphs have potential for regulatory approved medical devices" - has been put to me by many highly enthusiastic young techno-savvy entrepreneurial explorers of medical technology applications in the last months. All this and much more is addressed in this paper.

153

9 Comments

The growing field of genomics and precision medicine holds immense promise for advancing health equity, yet without addressing ethnic health disparities, it risks exacerbating existing inequities and leaving marginalized communities even further behind. In order to transform genomic medicine into a tool for social justice, ensuring benefits for all populations, we have to ensure inclusive research practices and equitable healthcare policies! #PrecisionMedicine #Genomics #Healthcare #HealthEquity

4

Thrilled to announce our latest publication in npj Digital Medicine titled “Computationally Derived Transition Points Across Phases of Clinical Care.” This study utilized statistical techniques to identify key transition points in patient care. By analyzing over 5.2 million patient encounters, we uncovered fundamental shifts in multimorbidity burdens across various ages using clustering analysis, False Nearest Neighbor, and transitivity analysis. Key Takeaways: 1. Redefining Clinical Phases - The study challenges the conventional age-based categorization of patients (e.g., <18, 18-65, >65). Our findings suggest more nuanced transitions that better reflect the underlying health complexities. 2. Early Adulthood Focus - The numerous transition points identified between ages 20 and 40 indicate significant health shifts during early adulthood. This period, often seen as clinically stable, shows diverse health changes that require tailored clinical attention. 3. Policy and Planning Implications - Health systems, payors, and policymakers can leverage these insights to better align healthcare utilization and resource allocation with true medical needs rather than relying solely on traditional age-based policies. 4. Future Research Directions: Further validation of these transition points in diverse patient populations and settings is needed to confirm these findings and refine our understanding of multimorbidity transitions. Read the full paper here: https://lnkd.in/eABC8drz A big thank you to all co-authors for their dedication and hard work! #Healthcare #AI #DigitalMedicine #Research #YaleUniversity #PatientCare #Multimorbidity #ClinicalTransitions Yale School of Public Health Yale Emergency Medicine Yale Biomedical Informatics & Data Science

30

1 Comment

UCSF to Develop AI Monitoring Platform for Clinical Care Platform aims to bridge the gap between the rapid evolution of AI technologies and the need for robust, ongoing assessment of their efficacy, safety, and equity

1

Circulatory ECMO is the way to bridge to definite therapy for difficult cases of high-risk acute PE's. Read our published review article in #Journal_CHESTCritCare.

16

Not all #AI is the same nor is all #healthcare AI the same. Insights from Zayed Yasin:

3

1 Comment