在人类标记数据上训练的人工智能工具反映了人类偏见:在一个大型临床连续膝关节关节炎队列中的案例研究
Anders Lenskjold1,2,3, Mathias W Brejnebøl4,5,6, Martin H Rose7
1Department of Radiology, Copenhagen University Hospital Bispebjerg-Frederiksberg, Copenhagen, Denmark. anders.lenskjold@regionh.dk.
Scientific reports
|November 5, 2024
概括
人工智能 (AI) 工具在对膝关节骨关节炎的分级上可能表现出不均性,类似于人类阅读器. 这项研究发现AI存在分歧,但与放射科医生相比,没有显著的准确性差异.
科学领域:
- 放射学 放射学是一门学科.
- 人工智能的人工智能
- 医疗成像医学成像
背景情况:
- 人类读者在医学图像解释中表现出偏见,导致疾病分级不统一.
- 在人类标记数据上训练的人工智能 (AI) 工具可能会继承这种不统一性.
- 评估AI性能需要评估潜在的偏见和不一致性.
研究的目的:
- 为了调查人工智能工具对膝关节骨关节炎的Kellgren-Lawrence分级的不均性.
- 将AI性能与高级放射科医生的性能进行比较.
- 在AI评分中识别潜在的年龄或性别偏见.
主要方法:
- 使用外部验证数据集 (50名患者) 和临床队列 (8273名患者).
- 在膝盖放射图上测试了一种FDA批准的AI工具,包括水平翻转的图像来评估侧面特定的不均性.
- 将AI准确度与高级放射科医生进行比较,并分析年龄和性别偏见.
主要成果:
- 人工智能工具显示出非统一性,在外部验证集上有20-22%的分歧,在队列上有13.6%.
- 人工智能工具和高级放射科医生之间没有发现精度的显著差异.
- 在AI工具对队列的表现中没有检测到年龄或性别偏见.
结论:
- 人工智能工具可以在医学图像分级中表现出不均性,反映出人类读者的变化.
- 虽然观察到AI的不均性,但它的性能仍然与专家放射科医生相美.
- 为了提高可靠性,需要对具有劣质AI性能的特定图像区域进行进一步调查.
相关概念视频
01:29Bias in Epidemiological Studies
161
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:
161
01:22Bias
3.7K
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...
3.7K