Data standards for the health information ecosystem have played a critical role in enabling software integration across health care enterprises for data sharing, analysis, clinical research, and public health. However, the ability to use large language models (LLMs) to dynamically extract unstructured data into a standardized form for downstream use poses a question about the future of health data. Namely, what role do data standards play in the era of LLMs, and will we need data standards at all? Gabriel Brat, MD, MPH, FACS, Josh Mandel, MD, and Matthew B.A. McDermott, PhD, address this question in a new editorial. Read “Do We Need Data Standards in the Era of Large Language Models?”: https://nejm.ai/4dcpyDb #HealthIT #AIinMedicine
NEJM AI’s Post
More Relevant Posts
-
By combining research with #AI and real-world data, scientists can improve clinical trial design, enrollment, and retention. Read this Q&A with Oracle Life Sciences’ Michael Fronstin to learn how platforms like Oracle Health Data Intelligence can make a difference. https://lnkd.in/e_R_dmwX
-
Course 2 - Introduction to Clinical Data Started off week 1 of the AI in healthcare with #StandfordMed Covered data mining and asking relevant questions. Data and research question types. Problems and biases that arise with the collation and analysis of data. Excited to learn more on Clinical Data.😊 #MedTech #AIinHealthcare #ClinicalData #HealthTech
-
-
By combining research with #AI and real-world data, scientists can improve clinical trial design, enrollment, and retention. Read this Q&A with Oracle Life Sciences’ Michael Fronstin to learn how platforms like Oracle Health Data Intelligence can make a difference. https://lnkd.in/exT8TiMT
How is Oracle technology helping CROs enable pharma to get trials up and running?
-
By combining research with #AI and real-world data, scientists can improve clinical trial design, enrollment, and retention. Read this Q&A with Oracle Life Sciences’ Michael Fronstin to learn how platforms like Oracle Health Data Intelligence can make a difference. https://lnkd.in/gjhxT7FQ
How is Oracle technology helping CROs enable pharma to get trials up and running?
-
Postdoctoral Researcher - Stanford Center for Artificial Intelligence in Medicine & Imaging (AIMI)
This is definitely an important contribution to the literature and delighted to be part of the SHC-Data Science Team Nigam Shah Nikesh Kotecha. Why? 1️⃣ This framework was developed by a multidisciplinary team with expertise in #AI/ML, digital health, IT, medicine, EHR, business management, and ethics—drawing from our collective experience at Stanford Health Care Technology & Digital Solutions - Stanford Medicine 2️⃣ Key Takeaway 🔑 Just because an AI solution exists doesn’t mean it will bring value to the patients we serve or improve clinicians' workflows. 3️⃣ We learnt so much about the implementation challenges and the critical role workflow design plays in determining AI's usefulness. I'm so honored and grateful for the amazing team #SHCDataScienceTeam Alison Callahan Duncan McElfresh Aditya Sharma Michael Pfeffer, Jonathan H. Chen, Dev Dash, Christopher Sharp, N. Lance Downing, MD , Anurang Revri, Nikesh Kotecha, Abby P., Srikar N., Jon Masterson, Conor K. Corbin, Michael Wornow, Rahul Thapa, Gabrielle Bunney, MD MBA, Danton Char, Michelle Mello, Aditya Sharma, Alaa Youssef Stanford University School of Medicine Stanford Center for Artificial Intelligence in Medicine and Imaging (AIMI) Stanford Biomedical Data Science Program
A testing and evaluation mechanism by the Stanford Data Science Team enables a process for accepting, reviewing, and supporting use case requests for artificial intelligence integration in an ethical and structured manner: https://nej.md/3zeENgn Stanford Health Care Alison Callahan Duncan McElfresh Juan M. Banda Nigam Shah
-
-
Pipeline Medical Affairs Director at Novo Nordisk. Public Health and Preventive Medicine Physician.
A terrific framework for more modeling in healthcare. Glad to see this published. 1. Define the problem: Y 2. Identify key determinants of the problem: Xn 3. Identify which variables to modify in order to solve the problem: f(x)=dY=Bo + B1*dX1 + …+ Bn*dXn 4. Prioritize the modifiable variables with the greatest expected change or impact (largest betas) based on your modeling Everything we do in healthcare can benefit from this, both conceptually and operationally.
A testing and evaluation mechanism by the Stanford Data Science Team enables a process for accepting, reviewing, and supporting use case requests for artificial intelligence integration in an ethical and structured manner: https://nej.md/3zeENgn Stanford Health Care Alison Callahan Duncan McElfresh Juan M. Banda Nigam Shah
-
-
Large Language Models have shown promise in analyzing unstructured clinical data to provide insights and recommendations. However, their complexity and opacity pose challenges in meeting the transparency requirements outlined in the HTI-1 Final Rule, particularly in terms of demonstrating the source of training data, identifying potential biases, and providing a clear understanding of how the algorithm arrived at its predictions. Read David Lareau’s latest blog to learn more: https://bit.ly/44Td00y | #HTI1 #Interoperability #ArtificialIntelligence #AIinHealthcare #AIGuardrails #LargeLanguageModels Office of the National Coordinator for Health Information Technology (ONC)
-
By combining research with #AI and real-world data, scientists can improve clinical trial design, enrollment, and retention. Read this Q&A with Oracle Life Sciences’ Michael Fronstin to learn how platforms like Oracle Health Data Intelligence can make a difference. https://lnkd.in/gwT4xztG
How is Oracle technology helping CROs enable pharma to get trials up and running?
-
Sometimes it’s good to go back to basics. A good video explaining fundamentals of machine learning (ML) particularly the point that in ML you do not need to define explicit rules. The beauty of the approach is that it learns the internal dynamics of a system using data.
Director of Innovation and Academic (Associate Professor) @ Edinburgh, Data Scientist in Healthcare, PhD (Oxford)
📃 Very happy to share that we recently published a correspondence in The Lancet that clarifies two distinct paradigms of modelling: a mechanistic approach and a statistical or machine learning-based approach. ➡ Our letter was a response to another correspondence that incorrectly assumed our model—reported in an earlier publication in the same journal—had a major limitation because it did not account for patients who may leave the NHS waiting list without treatment. 💡 This is a fascinating discussion to have in a premium clinical journal, but it’s easy for many practitioners in the field to get the nuances wrong. 💡Understanding the distinction between a mechanistic approach and a machine learning (or statistical) approach is vital, especially for those using computational techniques like #statistics, #datascience, or #machinelearning in #healthcare. ➡ While we addressed this in our correspondence, 400 words can’t fully capture such a fundamental concept. ➡ That’s why I’ve created a video explaining this distinction in under 9 minutes, where I hope you’ll see why the machine learning-based approach is so powerful for predictions. 👌 I’ve never been more convinced of its usefulness!
-
In the big scheme of healthcare operations, predictive models play a role not unlike that of blood tests, X-rays, or MRIs: They influence decisions about whether an intervention is appropriate. “Broadly speaking, models do math and yield probability estimates that help you decide whether to act,” says Nigam Shah, Chief Data Scientist for Stanford Health Care and a Stanford HAI faculty member. But those probability estimates are only useful to healthcare providers if they trigger decisions that are more beneficial than not. “As a community, I think we’re hung up on the performance of the model and not asking the question, Is the model useful?” Shah says. “We need to think outside the model.” Check out my blog post https://wix.to/i717sv9 #newblogpost #AIinHealthcare #HealthTech #MedicalAI #ArtificialIntelligence #HealthcareInnovation #DigitalHealth #AIforHealthcare #HealthcareTechnology #MachineLearning #BigData #PrecisionMedicine
Healthcare, Machine LearningHow Do We Ensure that Healthcare AI is Useful?
infinite-cure.com
Strategic Informatics Analyst @ Memorial Sloan Kettering Cancer Center | Responsible AI
2moHumility, compassion, and patience are human qualities no [model] can provide. But me thinks given the exorbitant cost in compute power and overall cost, applied thoughtfully and deliberately, large language models can certainly help our skilled human clinicians and organizations to focus on those qualities while improving health outcomes. Thoughts?