Our cross-sectional evaluation of six consumer-facing large language model platforms found significant heterogeneity in safeguard performance against the generation of health disinformation.
Read more: infodemiology.jmir.org/2026/…#LLMs#AI#Disinformation
In the largest analysis of individual participant data from trials supporting Food and Drug Administration drug approvals, we found that female patients had a 21% lower risk of death but a 12% higher risk of severe adverse events.
#metaanalysis#trialsdoi.org/10.1093/jnci/djag046
From the ANZ Alliance for Pancreas Research defining #AcutePancreatitis burden in #SouthAustralia
•High incidence vs global
•Mainly males, but increase in females
•Gallstones lead
•disease of disadvantage, rise in least disadvantaged areas
kwnsfk27.r.eu-west-1.awstrac…
AI scribing models can HEAR & SEE 👀
Vision-enabled AI scribes outperform audio-only models across medication histories - 98% vs 81% accuracy.
Fewer omissions. Safer documentation.
Full paper:
nature.com/articles/s41746-0…
Check out our latest paper
We built an AI scribe paired with RayBan Meta AI glasses. When put to the test with scribing medication histories - it was 98% accurate (+17% increase in accuracy compared to using audio alone)!
#AI#GenerativeAI#AIscribenature.com/articles/s41746-0…
Problem: Health disinformation chatbots
"OpenAI GPT Store were shown to be vulnerable to malicious system-level instructions to covertly create health disinformation chatbots."
Study: Malicious Conversion Into Health Disinformation Chatbots acpjournals.org/doi/10.7326/…