An ultrasound-based artificial intelligence framework for difficult airway prediction: A two-model, three-step decision framework
Chunmeng Fu, Cunyuan Luan, Huabo Liu, Wenfei Wang, Xia Zhou, Yuanfang Jia, Bing Ding, Lei Zhang, Li Yuan, Zejun Niu
Abstract
At present, the early warning of difficult airway remains fraught with challenges. Previous ultrasonic quantitative parameters have demonstrated favorable application potential in difficult airway assessment, and deep learning techniques have also exhibited satisfactory performance in the interpretation of this condition.
Introduction
Airway management is a core component of clinical practice in the intensive care unit (ICU), emergency department, and anesthesiology department [1,2]. In routine airway management, physicians across different departments exhibit variations in laryngoscope selection; the choice between direct laryngoscopes (DL) and video laryngoscopes (VL) is highly dependent on individual physicians’ clinical experience, and no unified standard protocol has been established [3].
Materials and methods
This prospective study was approved by the Ethics Committee of the Affiliated Hospital of Qingdao University (Approval No.: QYFYEC2024−68) and registered at the Chinese Clinical Trial Registry (Registration No.: ChiCTR2400084274). All study participants provided written informed consent.
Results
A total of 1693 patients undergoing general anesthesia were screened between May 2024 and April 2025. Among them, 134 were excluded due to age < 18 years, 15 due to language communication disorders, 202 due to emergency surgery, 268 due to non-tracheal intubation anesthesia.
Discussion
This study combines ultrasound imaging with deep learning techniques to predict difficult laryngoscopic exposure under DL and VL, and further proposes a two-model, three-step decision framework for perioperative airway risk stratification. In this framework, the CL-AI model showed good discrimination performance in identifying difficult DL exposure.
Conclusions
Based on the CNN deep learning algorithm, this study developed an AI classification model that integrates features derived from four ultrasound planes for predicting difficult laryngoscopic exposure, and further constructed a Two-Model, Three-Step airway management decision framework.
Citation: Fu C, Luan C, Liu H, Wang W, Zhou X, Jia Y, et al. (2026) An ultrasound-based artificial intelligence framework for difficult airway prediction: A two-model, three-step decision framework. PLoS One 21(2): e0342339. https://doi.org/10.1371/journal.pone.0342339
Editor: Chiara Lazzeri, Azienda Ospedaliero Universitaria Careggi, ITALY
Received: November 20, 2025; Accepted: January 21, 2026; Published: February 18, 2026
Copyright: © 2026 Fu et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data of this study cannot be shared publicly because the institutional review board prohibits public disclosure of subjects’ medical imaging data, and sensitive information such as time, gender, height, weight, and ultrasound images still has the risk of privacy disclosure even after deidentification. Data are archived in the hospital research project data management system, and researchers who meet the confidential data access standards can access the data through the Ethics Committee of the Affiliated Hospital of Qingdao University (qdfykygzb@163.com).
Funding: This study was financially supported by the Affiliated Hospital of Qingdao University (https://www.qduh.cn) in the form of a “Clinical Medicine + X Program” grant (QDFY+X202101057) received by ZN. This study was also financially supported by Jumei Ruiyan Pharmaceutical Technology Co., Ltd. in the form of a “2024 Jumei Ruiyan Anesthesia Research Fund Public Project” grant (Y2024080039) received by ZN. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
