Summary of Large Language Model Capabilities in Perioperative Risk Prediction and Prognostication, by Philip Chung et al.
Large Language Model Capabilities in Perioperative Risk Prediction and Prognostication
by Philip Chung, Christine T Fong, Andrew M Walters, Nima Aghaeepour, Meliha Yetisgen, Vikas N O’Reilly-Shah
First submitted to arxiv on: 3 Jan 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Computation and Language (cs.CL)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper explores the ability of general-domain language models, such as GPT-4 Turbo, to perform risk stratification and predict post-operative outcomes using clinical notes from electronic health records. The authors test the models’ predictive performance on eight tasks, including ASA Physical Status Classification, hospital admission, and mortality rates. They find that few-shot and chain-of-thought prompting can improve performance for certain tasks, achieving F1 scores of 0.50-0.86. While the models struggle with duration prediction tasks, they show promise in assisting clinicians with perioperative risk stratification on classification tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper looks at using big language models to help doctors predict patient outcomes after surgery. The models use information from patients’ electronic health records to make predictions about things like whether a patient will need to stay in the hospital longer or if they might die after the procedure. The researchers tested the models on eight different tasks and found that some of them did better than others. They also tried using different ways of giving instructions to the models, which helped improve their performance in some cases. Overall, this research suggests that big language models could be useful tools for doctors trying to figure out what might happen to patients after surgery. |
Keywords
» Artificial intelligence » Classification » Few shot » Gpt » Prompting
