使用大型语言模型在临床笔记中大规模识别偏差
Donald U Apakama1,2,3,4, Kim-Anh-Nhi Nguyen5, Daphnee Hyppolite6
1The Windreich Department of Artificial Intelligence and Human Health, Icahn School of Medicine at Mount Sinai, New York, NY.
Mayo Clinic proceedings. Digital health
|December 3, 2025
概括
生成性预训练变压器 (GPT) -4准确地检测和修改临床笔记中的偏见语言. 这种人工智能方法识别了导致文档偏差的可修改因素,可能减少医疗保健差异.
科学领域:
- 临床信息学 临床信息学
- 自然语言处理自然语言处理.
- 医疗保健的不平等 医疗保健的不平等
背景情况:
- 临床文档中偏见的语言可以使医疗保健差异持续存在.
- 识别和减轻这种偏见对于公平的患者护理至关重要.
研究的目的:
- 评估生成预训练变压器 (GPT) -4在急诊室 (ED) 检测和修改偏见语言的能力.
- 确定与临床笔记中偏见性文档相关的因素.
主要方法:
- 随机抽样采用了5万张ED账单和500张MIMIC-IV账单.
- GPT-4标记了四种类型的偏见:毁,耻辱/标签,判断和刻板印象.
- 人类审查员验证了GPT-4检测;多变量逻辑回归分析了偏差关联.
主要成果:
- 与人类审查相比,GPT-4在检测偏差方面表现出高灵敏度 (97.6%) 和特异性 (85.7%).
- 偏见的语言存在于ED笔记的6.5%和MIMIC-IV笔记的7.4%.
- 频繁的医疗保健使用,物质使用陈述和夜班与偏差增加有关;医生对GPT-4修订的评价很高.
结论:
- 在临床笔记中,GPT-4有效地检测并建议修改偏见的语言.
- 人工智能工具识别出可修改的对文档偏差的贡献者.
- 这项技术有望缓解偏见并减少医疗保健差异.
相关概念视频
Bias
7.2K
Bias refers to any tendency that prevents a question from being considered unprejudiced. In research, bias occurs when one outcome or answer is selected or encouraged over others in sampling or testing. Bias can occur during any research phase, including study design, data collection, analysis, and publication.
In statistics, a sampling bias is created when a sample is collected from a population, and some members of the population are not as likely to be chosen as others (remember, each member...
In statistics, a sampling bias is created when a sample is collected from a population, and some members of the population are not as likely to be chosen as others (remember, each member...
7.2K
Bias in Epidemiological Studies
1.2K
Biases can arise at various stages of research, from study design and data collection to analysis and interpretation. Recognizing and addressing these biases is essential to ensure the validity and reliability of epidemiological findings.Broadly speaking, biases in epidemiology fall into three main categories: selection bias, information bias, and confounding. A more detailed description of possible biases is:
1.2K
Improving Translational Accuracy
14.0K
Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...
14.0K
Improving Translational Accuracy
3.5K
3.5K
Biasing of FET
657
Biasing a Junction Field Effect Transistor (JFET) is crucial for setting operational parameters and ensuring efficient functioning in electronic circuits. JFETs are characterized by using a single carrier type in N-channel or P-channel configurations, where the channel is surrounded by PN junctions. These junctions are central to the device's ability to control current flow.
In an N-channel JFET, the structure consists of N-type material forming the channel on a P-type substrate, with the...
In an N-channel JFET, the structure consists of N-type material forming the channel on a P-type substrate, with the...
657
Motivational Bias
298
Cognitive bias results from limitations in thinking and information processing, leading to systematic errors in judgment. Conversely, motivational bias stems from personal desires or emotions, causing distortions in perception to align with self-interest. Motivational bias influences how individuals perceive and attribute causes to events, often shaped by personal needs, goals, and self-esteem preservation. This bias can distort judgment, leading to inaccurate assessments of success, failure,...
298


