Five AI models tackled 227 pharmacy exam questions. Here's where they struggled
Researchers evaluated five large language models using 227 infectious disease-related pharmacy examination questions and found that all achieved approximately 90% accuracy, with ChatGPT-4o-mini scoring highest at 91.2%. Accuracy was significantly lower for medicinal chemistry, clinical microbiology, higher-order cognitive questions, and image-containing questions, raising concerns about relying…
Read original article on The Medical News →Related news
-
Discontinuing statins generally safe for older adults without heart disease
United Kingdom
-
Intravenous antiplatelet therapy may improve heart function in cardiogenic shock patients compared to oral medication
United Kingdom
-
An inside look at a molecular switch that triggers inflammation
United Kingdom
-
Few meet major clinical response criteria after treatment escalation for lupus
United Kingdom
-
New blood test claims to detect 50+ cancers before symptoms—but there's no proof it reduces cancer death risk
United Kingdom
-
Scientists identify inhibition of a specific enzyme as a promising target to treat pancreatic cancer
United Kingdom
-
Researchers reveal how inhaled nanoplastics trigger lung inflammation
United Kingdom
-
Intensive blood pressure therapy did not cut cardiovascular events in adults with diabetes
United Kingdom