STEMI等危与模仿心电图导管室激活决策准确性:医师与人工智能(PMcardio的Queen of Hearts)对比
本研究纳入53名急诊医师和42名心内科医师,判读18份心电图(8份STEMI等危、8份STEMI模仿、1份STEMI和1份正常对照),判断是否存在需立即导管室激活的OMI,并与Queen of Hearts AI比较,参考标准基于血管造影、肌钙蛋白、超声心动图和临床随访。结果显示急诊医师与心内科医师准确率相似(65.6% vs 65.5%,p=0.969),均显著低于QoH AI(88.9% vs 医师总体65.6%,p<0.001)。医师最常误判de Winter、短暂性STEMI、超急性T波OMI和束支传导阻滞心电图;QoH AI仅误判左束支传导阻滞伴OMI和无OMI的左心室室壁瘤。结论提示AI有潜力减少漏诊OMI和不必要导管室激活,但需前瞻性研究验证。
要点
- 急诊医师与心内科医师判读STEMI等危和模仿心电图的准确率相似(65.6% vs 65.5%,p=0.969)。
- 医师总体准确率65.6%(95%CI 52-77)显著低于QoH AI的88.9%(95%CI 82-93),p<0.001。
- 医师最常误判de Winter、短暂性STEMI、超急性T波OMI和束支传导阻滞心电图。
- QoH AI仅误判左束支传导阻滞伴OMI和无OMI的左心室室壁瘤。
- 作者认为QoH AI有潜力减少漏诊OMI和不必要导管室激活,但需前瞻性研究验证。
适用人群与场景
- 53名急诊医师和42名心内科医师判读18份心电图(8份STEMI等危、8份STEMI模仿、1份STEMI和1份正常对照),判断是否存在需立即导管室激活的OMI。
研究设计与关键结果
- 研究类型为诊断准确性研究,纳入53名急诊医师和42名心内科医师,判读18份心电图(8份STEMI等危、8份STEMI模仿、1份STEMI和1份正常对照),参考标准基于血管造影、肌钙蛋白、超声心动图和临床随访。
- 急诊医师判读准确率65.6%(95%CI 51-78),心内科医师65.5%(95%CI 51-77),两者相似(p=0.969)。
- 医师总体准确率65.6%(95%CI 52-77)显著低于QoH AI的88.9%(95%CI 82-93),p<0.001。
- 医师最常误判de Winter、短暂性STEMI、超急性T波OMI和束支传导阻滞心电图;QoH AI仅误判左束支传导阻滞伴OMI和无OMI的左心室室壁瘤。
证据类型、等级与局限
- 证据类型为诊断准确性研究。
- 局限:原文未报告局限性。
对床旁实践的改变
- 原文结论指出医师常误判STEMI等危和模仿心电图,可能影响导管室激活决策;QoH AI显示出更高准确率,作者认为有潜力减少漏诊OMI和不必要导管室激活。
- 原文指出需前瞻性研究验证这些发现。
提炼说明
- 本条依据论文摘要整理;摘要未涉及「诊断标准与关键阈值」「处置流程与药物剂量」「红旗征与不能漏的情况」「指南或共识出处」,故未列出这些栏目;全文请点「阅读原文」。
英文原文(摘要)· 点击展开
INTRODUCTION: Accurate ECG interpretation is crucial to identify occlusive myocardial infarction (OMI) to determine the need for immediate catheterization laboratory activation (CLA). STEMI-equivalent and STEMI-mimic ECG patterns deviate from conventional STEMI criteria, risking misclassification of OMI cases. The diagnostic accuracy for these complex ECGs is unknown.
OBJECTIVES: This study aimed to measure physician accuracy for interpreting STEMI-equivalent and STEMI-mimic ECGs for catheterization laboratory activation (CLA) and compare their performance to a machine learning-based artificial intelligence algorithm, Queen of Hearts AI (QoH AI).
METHODS: Fifty-three EPs and 42 cardiologists interpreted 18 ECGs (eight STEMI-equivalents, eight STEMI-mimics, with one STEMI, and a normal ECG as controls) to determine the presence of OMI requiring immediate CLA. The same ECGs were analyzed by QoH AI. Interpretations were compared against a reference standard based on angiography, troponin, echocardiography, and clinical follow-up.
RESULTS: Interpretation accuracies were similar between EPs and cardiologists (65.6 %, 95 % CI [51, 78]; 65.5 %, 95 % CI [51, 77], respectively; p = 0.969), and significantly lower than QoH AI (88.9 %, 95 % CI [82, 93]) vs. physicians overall, 65.6 %, 95 % CI [52, 77]; p < 0.001). Physicians most frequently misclassified de Winter, Transient STEMI, Hyperacute T-wave OMI, and bundle branch block ECGs. QoH AI only misclassified left bundle branch block with OMI and left ventricular aneurysm without OMI.
CONCLUSION: Physicians frequently misinterpret STEMI-equivalent and STEMI-mimic ECGs, potentially impacting CLA decisions. QoH AI demonstrated superior accuracy, suggesting a potential to reduce missed OMIs and unnecessary catheterization laboratory activations. Prospective studies are needed to validate these findings in clinical practice.
参考来源
- Shroyer S, Mehta S, Thukral N, et al. Accuracy of cath lab activation decisions for STEMI-equivalent and mimic ECGs: Physicians vs. AI (Queen of Hearts by PMcardio). Am J Emerg Med. 2025. PMID: 40763602. doi:10.1016/j.ajem.2025.07.061
供执业医师抢救与诊疗参考,不替代临床判断;剂量请结合体重、肝肾功能和本院药品规格复核,以指南原文为准。