Comparing weight-loss medications; using AI to localize seizures; can a virtual nurse convince patients to get vaccinated? – Morning Medical Update

Author(s):

Austin Littrell

Fact checked by:

Keith A. Reynolds

Key Takeaways

Tirzepatide led to greater weight loss than semaglutide, with participants losing 20.2% of body weight on average in the SURMOUNT-5 trial.
ChatGPT matched or outperformed epileptologists in predicting common seizure zones, but was less accurate in rarer regions.
A virtual nurse, FLORA, increased vaccination willingness by 33 times when paired with direct recommendations, highlighting AI's potential in healthcare interventions.

The top news stories in medicine today.

Tirzepatide outperforms semaglutide in head-to-head obesity trial

In a direct comparison of two leading GLP-1-based weight loss medications, tirzepatide (Zepbound) led to significantly greater weight loss than semaglutide (Wegovy), according to results from the 72-week SURMOUNT-5 phase 3b trial published in the New England Journal of Medicine. Participants with obesity but no diabetes lost an average of 20.2% of their body weight — roughly 50 pounds — on tirzepatide, compared to 13.7% (33 pounds) on semaglutide. The trial, funded by Eli Lilly, found that nearly one-third of tirzepatide users lost at least 25% of their body weight, double the rate seen with semaglutide. Side effects were similar across both groups, with nausea and abdominal pain among the most common. Researchers attribute tirzepatide’s stronger effect to its dual action on GIP and GLP-1 receptors, and note that next-generation drugs, like “triple G” retatrutide, are already in development.

ChatGPT helps locate seizures, outperforming experts

Researchers at Stevens Institute of Technology found that ChatGPT could assist neurosurgeons by accurately predicting the brain regions responsible for epileptic seizures based on seizure symptom descriptions. In a comparison study with five board-certified epileptologists, ChatGPT matched or outperformed human experts in common seizure zones like the frontal and temporal lobes. However, it was less accurate in rarer regions, like the insula. To improve precision, the team developed a specialized large language model, “EpiSemoLLM,” trained specifically on seizure semiology. The study, published in Journal of Medical Internet Research, highlights the potential for human-AI collaboration to improve pre-surgical assessments for epilepsy treatment.

Can a virtual nurse convince patients to get vaccinated?

A virtual nurse named FLORA significantly increased participants’ willingness to get vaccinated against the flu, especially when paired with a direct recommendation, according to a new study from SWPS University. Researchers tested the tool on nearly 1,800 participants and found that virtual dialogue personalized to a user’s health status — delivered via an empathetic on-screen avatar — was more persuasive than risk assessments or printed materials alone. When FLORA’s interaction included a clear request to get vaccinated, the willingness to get a flu shot jumped by 33 times.