Summary of Humans Continue to Outperform Large Language Models in Complex Clinical Decision-making: a Study with Medical Calculators, by Nicholas Wan et al.
Humans Continue to Outperform Large Language Models in Complex Clinical Decision-Making: A Study with Medical Calculators
by Nicholas Wan, Qiao Jin, Joey Chan, Guangzhi Xiong, Serina Applebaum, Aidan Gilson, Reid McMurry, R. Andrew Taylor, Aidong Zhang, Qingyu Chen, Zhiyong Lu
First submitted to arxiv on: 8 Nov 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
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 Medium Difficulty summary: The paper investigates the ability of large language models (LLMs) and medical trainees to recommend medical calculators in various clinical scenarios, such as risk stratification, prognosis, and disease diagnosis. The authors evaluated eight LLMs, including open-source, proprietary, and domain-specific models, using 1,009 question-answer pairs across 35 clinical calculators. They found that while the highest-performing LLM, GPT-4o, achieved an accuracy of 74.3%, human annotators outperformed LLMs with an average accuracy of 79.5%. The study highlights that humans continue to surpass LLMs on complex clinical tasks due to comprehension and calculator knowledge errors. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Low Difficulty summary: This paper looks at how well artificial intelligence (AI) language models and medical students can choose the right medical tools for different patient situations. Researchers tested eight AI models using 1,009 questions about 35 different medical tools. They found that while one AI model did pretty well, getting 74% of answers correct, human experts still got more answers right on average. The study shows that humans are still better at making decisions in complex medical situations. |
Keywords
» Artificial intelligence » Gpt
