使用大型语言模型提取成本效益分析数据:一个案例研究
Xujun Gu1, Hanwen Zhang2, Divya Patil3
1Department of Practice, Sciences, and Health Outcomes Research, University of Maryland Baltimore, Baltimore, MD, USA.
大型语言模型 (LLM) 显示出用于自动化成本效益分析 (CEA) 数据提取的承诺,实现与已建立的注册表可比的准确性. 人类监督对于复杂的数据点至关重要.
科学领域:
- 卫生经济学 卫生经济学
- 医疗保健中的人工智能
- 医疗信息学医学信息学
背景情况:
- 成本效益分析 (CEA) 对健康经济研究至关重要,但依赖于手动数据收集,这是劳动密集型和易出错的.
- 人工智能 (AI) 和大型语言模型 (LLM) 的进步为CEA提供了自动化数据提取的机会.
研究的目的:
- 评估基于LLM的数据提取对CEA的准确性.
- 评估使用LLM来支持CEA数据收集的可行性.
主要方法:
- 与Tufts CEA注册表 (TCRD) 和研究人员验证数据 (RVE) 进行了比较,使用自定义的ChatGPT模型 (GPT).
- 这些模型从34个结构文章中提取了36个预定义的变量.
- 计算和统计分析了对应率和准确度差异.
主要成果:
- GPT的准确性与TCRD相当 (GPT与RVE相比:平均0.88,TCRD与RVE相比:平均0.90).
- 在提取"人口和干预细节"方面,GPT表现出色,但在处理"效用"等复杂变量方面面临挑战.
结论:
- 像GPT这样的LLM显示出自动化CEA数据提取的潜力,提供与现有方法相似的准确性.
- 人类监督对于管理复杂性和确保准确性至关重要,特别是对于复杂的变量.
更多相关视频
07:05Operant Protocols for Assessing the Cost-benefit Analysis During Reinforced Decision Making by Rodents
Published on: September 10, 2018
05:47Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems
Published on: June 13, 2025
相关概念视频
Methods of Documentation VI: Case Management Model
For example, a patient with a chronic...
Mechanistic Models: Compartment Models in Individual and Population Analysis
Types of Biopharmaceutical Studies: Controlled and Non-Controlled Approaches
Non-controlled studies, commonly employed for initial exploration, lack a control group, rendering them susceptible to biases and external influences. In contrast,...
Analysis Methods of Pharmacokinetic Data: Model and Model-Independent Approaches
The model approach uses mathematical models to describe changes in drug concentration over time. Pharmacokinetic models help characterize drug behavior in patients, predict drug concentration in the body fluids, calculate optimum dosage regimens, and evaluate the risk of toxicity. However, ensuring that the model fits the experimental data accurately...
Analysis of Population Pharmacokinetic Data
Statistical Methods for Analyzing Epidemiological Data
