Authors: Liu H et al.
Source: Anesthesia & Analgesia. December 2025. Volume 141(6):1435–1437. DOI: 10.1213/ANE.0000000000007587
Summary:
In this feasibility pilot study, Liu and colleagues explore whether robotic automation combined with artificial intelligence can safely and effectively perform flexible optical bronchoscope (FOB) airway insertion. Given the technical difficulty of FOB intubation, the dependence on operator skill, and declining first-attempt success in less experienced hands, the authors sought to determine whether a robotic system could standardize performance and reduce error.
The investigators developed the Robotic Flexible Intubation System (RFIS), a master–slave robotic platform integrating a flexible videoendoscope, a three-degree-of-freedom manipulator, and a human–machine interface powered by a convolutional neural network for real-time airway anatomy recognition and navigation. RFIS can operate in two modes: robot-assisted (human-guided) and fully automated. Performance was tested in a standardized airway manikin by lay operators with no prior bronchoscopy experience and compared against unassisted manual FOB insertion by experienced anesthesiologists.
First-attempt success was the primary outcome. Automated RFIS-guided insertion by lay users achieved a 100% first-attempt success rate, while robot-assisted insertion achieved 92.5%, both markedly superior to unassisted manual insertion by lay users (62.5%) and comparable to experienced anesthesiologists (95%). Performance quality scores followed a similar pattern: fully automated RFIS insertion achieved perfect median scores equivalent to anesthesiologists and substantially higher than unassisted lay performance. Insertion times were longer with RFIS compared to experienced anesthesiologists but remained clinically reasonable, suggesting current limitations were more related to system hardware and processing speed than feasibility.
The authors conclude that automated FOB airway insertion using RFIS is technically feasible and, in a controlled simulation environment, can match expert-level performance even when operated by nonclinicians. While substantial limitations remain—including manikin-only testing, absence of tube placement, lack of nasal insertion capability, and challenges related to secretions, patient movement, and awake airways—the work provides proof of concept that artificial intelligence–driven robotic airway navigation may one day augment or extend anesthetic airway care in select scenarios.
What You Should Know:
• Flexible bronchoscopy is highly operator dependent, with lower first-pass success among less experienced clinicians.
• A robotic system using AI-based airway recognition achieved expert-level success and quality in a simulation model.
• Fully automated airway navigation performed as well as experienced anesthesiologists in this pilot study.
• Translation to real patients will require advances in tube placement, secretion handling, patient motion compensation, and safety validation.
Key Points:
• Automated RFIS achieved 100% first-attempt success in a manikin model operated by lay users.
• Robot-assisted and automated modes significantly outperformed unassisted manual FOB insertion by nonexperts.
• Performance quality with automated RFIS matched that of experienced anesthesiologists.
• This study establishes feasibility but not clinical readiness for robotic airway insertion.
Thank you to Anesthesia & Analgesia for publishing this provocative pilot study that pushes the boundary of airway management toward artificial intelligence, robotics, and future automation.