Complications like pneumonia or infections can ruin recovery after surgery. Over 10% of patients face these issues. They often stay longer in the ICU, face higher death risks, and raise medical costs. Spotting at-risk patients early matters a lot. Yet, guessing these risks accurately is hard.

A new study brings hope with artificial intelligence (AI). Researchers at Washington University have used large language models (LLMs) to study patient notes before surgery to predict problems after. A study published in npj Digital Medicine proves these LLMs beat older methods.

Surgery has big risks. Clinical notes hold key details from doctors. The new LLM, built just for surgery notes, predicts issues early. This lets doctors act fast, keeping patients safer. Traditional ways use structured data, like lab results or patient age. These miss the full story notes tell.

The researchers train their LLM on medical texts and health records. Then, they tweak it with surgery notes. This new method reads notes to spot risks better than old tools. It finds patterns in patient health that others skip.

AI could make surgery safer for everyone

The researcher tested this on 85,000 surgery notes from 2018 to 2021. The model caught 39 more complication cases per 100 than older AI. This could help doctors stop problems early. The study also shows foundation models, versatile AI that handles many tasks, work great here.

In a Washington University press release, the researchers note that these models adapt well. They predict lots of complications at once. Since issues often link together, this shared approach beats single-focus models. Experts see this tool fitting many hospitals. It could warn doctors about risks soon enough to act.

This AI method promises better care. By reading notes, it spots dangers fast. Doctors can then plan smarter fixes. The study marks a step toward using AI to make surgery safer for everyone.

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter.