Some writing tips

Large language models (#LLMs) have the potential to improve healthcare, but they could also be misused. This perspective from authors at NIH Clinical Center discusses how LLMs may favor system-wide benefits over individual rights, raising concerns about AI-enabled social credit

Electronic Health Record (EHR) data in pediatrics often lack detail, making data analysis challenging. This article introduces a transfer learning approach to enhance pediatric EHR analysis by creating specialized code embeddings that combine data from both pediatric and

Real-time intraoperative brain surgery with a visual foundation #AI model: In less than 10 seconds, accurate detection of tumor (glioma) infiltration from fresh surgical tissue without labels, validated in a prospective multi-center trial, made open-source nature Todd Hollon

📢 Seeking #HealthTech partners! Need: Organizations who can: Link medical imaging + path reports Share de-identified data sets Reach out by DM / email / linkedin

How to write an excellent Review article Nature Reviews Bioengineering nature #PhD #PhDlife #academics #AcademicChatter #AcademicTwitter #Postdocs #Research #Science

Large language models (LLMs) are typically optimized to answer peoples’ questions. But there is a trend toward models also being optimized to fit into agentic workflows. This will give a huge boost to agentic performance! Following ChatGPT’s breakaway success at answering

DataLLM - LLMs that generate SQL and operate on your data is becoming a bigger use case than RAG

There is a lot of energy going into fine-tuning models, but specialized medical AI models lost to their general versions 38% of the time, only won 12%. Before spending millions on specialized training, might be worth exploring what base models can do with well-designed prompts.

Remember exercise pages from textbooks? Large-scale collection of these across all realms of knowledge now moves billions of dollars. Textbooks written primarily for LLMs, compressed to weights, emergent solutions served to humans, or (over time) directly enacted for automation.

New NEJM and #AHA24 A large randomized trial in people with Type 2 diabetes showed intensive BP lowering (systolic < 120 mm Hg) reduced major cardiovascular outcomes nejm.org/doi/full/10.10…

CRISPR genome editing clicks for amyloid cardiomyopathy. New NEJM #AHA24 nejm.org/doi/full/10.10… A Phase 1 study with marked reduction of serum TTR marianna fontana UCL

The “personality” and “opinions” of AI are not stable, they are influenced by prompting & sycophancy Ab example: when GPT-4 was prompted in Korean to act Korean & prompted in English to act American, GPT-4 replicated Big-5 personality differences between Koreans & Americans

AWS just released a new Multi-Agent AI framework It lets you manage multiple AI agents, dynamically route LLM queries, maintain context across AI Agents and can be deployed locally on your computer. 100% opensource.

NEW Viewpoint: Building robust, proportionate, and timely approaches to regulation and evaluation of digital mental health technologies. MHRAgovuk NICE Wellcome Read it here: buff.ly/3UXiX93

A new foundation model for nearly all (9!) medical imaging modalities, BiomedParse nature.com/articles/s4159… Nature Methods Microsoft Research Hoifung Poon Sheng Wang University of Washington Explainer microsoft.com/en-us/research…

In 23 countries, among >87,000 people age 50+, internet use was associated with better mental health (not so much in the US) nature.com/articles/s4156… Nature Human Behaviour

Jensen Huang says AI Scaling Laws will increase computing power by a million times in the next decade, compared to Moore's Law which increased at 100 times per decade

I am recruiting PhD students at Duke! Please apply to Duke CompSci or Duke CBB if you are interested in developing new methods and paradigms for NLP/LLMs in healthcare. For details, see here: monicaagrawal.com/home/research-…. Feel free to retweet!

Fascinating chart from The Economist. It says the overall difficulty in learning a new language depends on how different that language is from your own. This chart is based on the US State Department's framework for its diplomats. Someone should do a similar study for Indians