"All tested LLMs performed poorly on medical code querying, often generating codes conveying imprecise or fabricated information. LLMs are not appropriate for use on medical coding tasks without additional research."
"We compare per capita ratios with an approach based on regression, a widely used statistical procedure that eliminates many of the problems with ratios and allows for straightforward data interpretation."
"With stronger freshwater anomalies, our results indicate an increase in the risk of warm, dry European summers and of heat waves and droughts accordingly."
Join APDU on January 17th for a discussion of the Do No Harm Guide: Collecting, Analyzing, and Reporting Gender and Sexual Orientation Data with Jonathan Schwabish, Urban Institute
Census Bureau experts recommend NOT using the Post Enumeration Survey (PES) to update the 2020 population estimates because "While the PES is helpful in identifying coverage issues at the national level, it is not able to identify them as accurately at lower levels of geography because of its design. The PES sample size was simply too small."
Today, a young American woman between the ages of 25 and 34 face higher mortality rates than at any other point in more than 50 years. And had the mortality rate remained flat between 2000 and 2021, nearly 40,000 young women would not have died.
~Sara Srygley of PRB
📣 Calling all Census afficionados:
Here's your chance to help shape the 2030 Census!
The Census Bureau is requesting nominations of members representing stakeholder organizations, groups, interests, and viewpoints to the 2030 Census Advisory Committee.
@meredithw@bibliolater@science Good point, as a medical layperson I would have not thought of that. Since the causal interpretation seems mechanistically plausible, it is tempting. I have yet to read the full paper but its abstract and conclusion are more carefully phrased than the reporting about the paper. I wonder if one can factor out negative correlation somehow in study design or analysis.