2025 年第 10 期 第 20 卷

人工智能在预测困难气道方向的研究进展

Research progress of artificial intelligence in predicting difficult airway

作者：杨展眉唐之音刘洪涛

英文作者：Yang Zhanmei Tang Zhiyin Liu Hongtao

单位：中国医科大学附属盛京医院麻醉科，沈阳110000

英文单位：Department of Anesthesiology Shengjing Hospital of China Medical University Shenyang 110000 China

关键词：人工智能；麻醉；气道管理；预测模型；机器学习；深度学习

英文关键词：

• 摘要：

• 气道管理是麻醉中重要的环节，难以预料的困难气道可以导致严重的后果，如低氧血症、脑死亡等，因此麻醉前的气道评估尤为重要。机器学习和深度学习在建立困难气道预测模型方面有着比传统方法（如改良马氏分级、甲颏距离等）更加优越的性能。机器学习算法可以敏锐发现变量之间的非线性关系，深度学习算法，特别是卷积神经网络及其变体，通过自动分析多模态数据，显著提高预测准确性和效率。本文重点介绍机器学习与深度学习技术在困难气道评估方面的应用，以期提高麻醉的安全性。

• Airway management is an important part of anesthesia. Unpredictable difficult airway can lead to serious consequences, such as hypoxemia and brain death, so airway assessment before anesthesia is particularly important. Machine learning and deep learning have better performance than traditional methods  (such as modified Maholobian classification, thyromental distance, etc.) in establishing difficult airway prediction models. Machine learning algorithms can keenly find nonlinear relationships between variables. Deep learning algorithms, especially convolutional neural networks and their variants, have significantly improved prediction accuracy and efficiency by automatically analyzing multimodal data. This article focuses on the application of machine learning and deep learning technology in difficult airway assessment, in order to improve the safety of anesthesia.