🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Of over 500 LLM #AI reports in healthcare, only 5% used real patient data. jamanetwork.com/journals/jama/… JAMA Nigam Shah

Congrats to Christopher A. Longhurst, MD for being awarded the AMIA Don Eugene Detmer for his health policy advocacy work in informatics! We are lucky to have his leadership at UC San Diego Health, where we can test and shape informatics policy ideas in practice. amia.org/news-publicati…

New research out now! 🍔🏫 We mapped food outlets around schools in regional & metro Australia using Google Street View, finding that unhealthful options have dominated for 17 years. Time for action to promote healthier environments for adolescents! 👇 onlinelibrary.wiley.com/doi/full/10.10…

A Johns Hopkins University study found that measuring 162 quality metrics cost >$5 mil. Our team at UC San Diego Health, led by Drs. Aaron Boussina and Shamim Nemati, found that LLMs are a great way to scale quality measurement - needed for a learning health system. Paper: ai.nejm.org/stoken/default…

Modernizing Clinical Research Data jamanetwork.com/journals/jama/…

Researchers developed a new methodology for identifying fake papers "which automatically tracks and groups suspicious figures and documents." journals.plos.org/plosone/articl…

Misinformation on social media leads to ineffective voting decisions psypost.org/misinformation… "...a laboratory experiment to simulate how social media influences voting decisions..." "...misinformation can drive users to disengage from information-sharing platforms, which might,

🎯 Thrilled to share our latest research published in Nature Communications! We've developed TrialGPT - an AI framework that helps match patients with clinical trials up to 43% more effectively than previous methods. 🔬 Key breakthroughs: - 90% accuracy in identifying relevant

Join experts from Japan and Australia as they explore how AI is transforming healthcare, enhancing diagnostics, personalising treatments, and improving patient care. Gain valuable insights into shaping the future of medicine. Don't miss this opportunity to learn from leaders in

Really interesting paper from Carol El-Hayek on an iterative ML approach in EHR data. We try to find people who inject drugs when the standard rules only find a biased subset and we don't have enough information about nearly everyone else nature.com/articles/s4174…

1/ I'll be at #NeurIPS2024 presenting our work SmallToLarge (S2L): Data-efficient Fine-tuning of LLMs! 🚀 What’s S2L? It’s a scalable data selection method that trains a small proxy model to guide fine-tuning for larger models, reducing costs while preserving performance. 👇

New Collection Alert! 🎯Multimodal AI for Digital Medicine Deadline: 17 September, 2025. Edited by Jinman Kim, Adam Dunn, and Lei Bi. nature.com/collections/aj…

dphconf.org/calls-for-pape… Digital Public Health conference, check the very interesting topics

Launching npj Digital Public Health: nature.com/npjdigipubheal… Research on development, implementation, & application of new digital methods & tools in public health; community, population, & global health, all diseases, & cultural & socioeconomic determinants of health.

sydney.edu.au/news-opinion/n…

📢📜 Excited to share that our paper "On the conversational persuasiveness of GPT-4" has been published in Nature Human Behaviour! 🤖 Key takeaway: LLMs can already reach superhuman persuasiveness, especially when given access to personalized information nature.com/articles/s4156…

2. SubPOP led by Joseph Jeesung Suh - we show that fine-tuning LLMs on high-quality survey data greatly improves their ability to represent diverse opinions of different subpopulations and generalizes to unseen surveys and groups! - arxiv.org/abs/2504.07114 - arxiv.org/abs/2502.16761

We focus so much on what genAI can do. This paper asks what users can make them do. Steerability -- how easily users can guide a model to produce what they want, is a missing dimension in genAI evaluation, and this paper shows how to measure it. arxiv.org/abs/2503.17482

At MESH Incubator™ at Mass General Brigham 2025, Harvard Medical School’s Isaac Kohane shared a striking case: GPT-4 recommended treatment to a patient when assessing from the perspective of a doctor—but denied treatment as an insurer. Same patient, same facts, different values. #Harvard harvardmagazine.com/2025/05/harvar…