文献速递胸部外周血管技术与安全CT

使用大型语言模型对FDA批准的人工智能进行部署后监测:肺栓塞检测应用案例

Using a Large Language Model for Postdeployment Monitoring of FDA-Approved Artificial Intelligence: Pulmonary Embolism Detection Use Case.
PubMed 文献 · J Am Coll Radiol · 2025-06-30

该研究回顾性分析11,999例CT肺动脉造影检查,构建了结合大型语言模型免费文本分类与人工审核的部署后监测框架,用于监测商业肺栓塞检测人工智能(CVPED)的性能。框架以连续7天差异率超过固定95%置信区间定义漂移,最佳回顾评估期为2个月,可检测模拟的敏感度与特异度下降并发现增量肺栓塞病例。

要点

  • 在11,999例CT肺动脉造影中,1,285例(10.7%)有肺栓塞,373例(3.1%)CVPED与大型语言模型分类不一致。
  • 模拟大型语言模型特异度下降2.5%导致CVPED阴性而大型语言模型阳性差异增加1.7倍,所提框架需22天检出。
  • 框架从111例CVPED阳性而大型语言模型阴性病例中识别出4例增量肺栓塞真警报。

核心结论

  • 结合基于大型语言模型的免费文本分类与人工在环警报系统的部署后监测框架,可连续追踪基于图像的AI性能、预警性能漂移并提供增量临床价值。

研究设计与样本

  • 回顾性分析11,999例于2023年4月30日至2024年6月17日期间进行的CT肺动脉造影检查;以大型语言模型放射报告分类联合CVPED输出并经人工审核分歧作为参考标准;模拟每日监测框架追踪CVPED与大型语言模型间分歧;通过系统改变CVPED或大型语言模型敏感度与特异度模拟漂移;纳入人工在环选择性警报框架进行连续前瞻性评估。

诊断标准与关键数值

  • 漂移定义为差异率连续7天超过固定95%置信区间;该置信区间与最佳回顾评估期由性能稳定的一致数据集确定,最佳回顾评估期确定为2个月;大型语言模型特异度下降2.5%导致CVPED阴性而大型语言模型阳性差异增加1.7倍,所提框架需22天检出;模型特异度下降2%至3%导致差异增加2至3倍,敏感度需下降10%才产生类似效应。

临床应用与阅片要点

  • 该框架可用于持续监测已部署的肺栓塞检测AI性能,通过人工在环选择性警报调查潜在增量检出;研究显示放射科医师未与CVPED交互的病例中,111例CVPED阳性而大型语言模型阴性病例有29例触发警报,其中24例为CVPED假阳性、1例为大型语言模型假阴性,最终识别4例增量肺栓塞真警报。

局限性

  • 原文未报告局限性

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

英文原文(摘要)· 点击展开

BACKGROUND: Artificial intelligence (AI) is increasingly integrated into clinical workflows. The performance of AI in production can diverge from initial evaluations. Postdeployment monitoring (PDM) remains a challenging ingredient of ongoing quality assurance once AI is deployed in clinical production.

PURPOSE: To develop and evaluate a PDM framework that uses large language models (LLMs) for free-text classification of radiology reports, and human oversight. We demonstrate its application to monitor a commercially vended pulmonary embolism (PE) detection AI (CVPED).

METHODS: We retrospectively analyzed 11,999 CT pulmonary angiography studies performed between April 30, 2023, and June 17, 2024. Ground truth was determined by combining LLM-based radiology report classification and the CVPED outputs, with human review of discrepancies. We simulated a daily monitoring framework to track discrepancies between CVPED and the LLM. Drift was defined when discrepancy rate exceeded a fixed 95% confidence interval for 7 consecutive days. The confidence interval and the optimal retrospective assessment period were determined from a stable dataset with consistent performance. We simulated drift by systematically altering CVPED or LLM sensitivity and specificity, and we modeled an approach to detect data shifts. We incorporated a human-in-the-loop selective alerting framework for continuous prospective evaluation and to investigate potential for incremental detection.

RESULTS: Of 11,999 CT pulmonary angiography studies, 1,285 (10.7%) had PE. Overall, 373 (3.1%) had discrepant classifications between CVPED and LLM. Among 111 CVPED-positive and LLM-negative cases, 29 would have triggered an alert due to the radiologist not interacting with CVPED. Of those, 24 were CVPED false-positives, 1 was an LLM false-negative, and the framework ultimately identified 4 true-alerts for incremental PE cases. The optimal retrospective assessment period for drift detection was determined to be 2 months. A 2% to 3% decline in model specificity caused a 2- to 3-fold increase in discrepancies, and a 10% drop in sensitivity was required to produce a similar effect. For example, a 2.5% drop in LLM specificity led to a 1.7-fold increase in CVPED-negative LLM-positive discrepancies, which would have taken 22 days to detect using the proposed framework.

CONCLUSION: A PDM framework combining LLM-based free-text classification with a human-in-the-loop alerting system can continuously track an image-based AI's performance, alert for performance drift, and provide incremental clinical value.

来源:J Am Coll Radiol,查看原文。版权归原作者所有。

引用

  1. Sorin V, Korfiatis P, Bratt AK, et al. Using a Large Language Model for Postdeployment Monitoring of FDA-Approved Artificial Intelligence: Pulmonary Embolism Detection Use Case. J Am Coll Radiol. 2025. PMID: 40602699. doi:10.1016/j.jacr.2025.06.036

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。