AI Laryngoscopy vs. Traditional Video: Top 3 Differences

The Glottic Co-Pilot

Video laryngoscopy (VL) shifted airway management by taking our eyes off the mouth and putting them on a screen. The evidence for that shift is now strong. In the DEVICE trial (NEJM, 2023), 1,417 critically ill adults intubated in EDs and ICUs had first-attempt success of 85.1% with VL vs 70.8% with direct laryngoscopy, with no difference in severe complications. In the field, a national cohort of 22,132 intubated cardiac arrest patients found VL associated with higher first-pass success (75.1% vs 69.5%), though not with more ROSC.

But traditional VL still depends entirely on the operator's spatial judgment under stress. A screen shows you the anatomy; it does not tell you what you are looking at. An emerging class of software tries to do that, and most of it is still in trials rather than mounted on shipping rigs. Here are the three differences that matter, and how far along each one actually is.

1. Real-Time Anatomical Detection

• Traditional VL: shows a camera feed. If secretions, blood, or fog obscure the lens, or trauma distorts the anatomy, the operator has to work it out.
• AI-guided VL: a computer-vision model labels landmarks on the live image. In a 2025 study from China, a YOLO-based model trained on images from 978 patients identified the glottic fissure with 94% precision but the aryepiglottic folds with only 62%. Deployed on a smartphone with a video laryngoscope, it cut simulated intubation time for novice trainees by 11 seconds and raised first-attempt success by 19.4%.
• Where it is: research and simulation, with one notable exception below.

2. Tube-Position Feedback

• Traditional VL: you watch the tube pass the cords and then confirm with waveform capnography.
• AI-guided VL: larynGuide, software that runs on the KARL STORZ C-MAC, gives audio-visual guidance during laryngoscopy and advises on whether the tube went into the trachea. In its first in-human study (110 elective surgical patients, software hidden from the operator), it correctly identified 102 of 108 tracheal placements, a sensitivity of 94%. It caught only one of two esophageal intubations, and fogging and poor visibility degraded its performance.
• In simulation: a randomized crossover trial with 74 clinicians found first-attempt success of 89% with the larynGuide overlay vs 85% without, at the cost of a longer median time to intubation (29 s vs 23 s).
• Where it is: early clinical. Low specificity for esophageal placement is exactly the failure mode we can't afford, which is why waveform EtCO2 stays the confirmation standard no matter what the screen says.

3. Force and Technique Feedback

• Traditional VL: the blade gives no feedback. Excessive leverage against the upper teeth or soft tissue goes unmonitored.
• Sensorized VL: researchers have fitted video laryngoscopes with force and position sensors that give sound and visual feedback when a force threshold is exceeded. So far that has been a training tool on manikins, not a feature on clinical devices.
• Where it is: simulation and education.

The Practical Takeaway

AI-guided blades don't pass the tube for you, and none of them has outcome data in the field yet. The plausible near-term win is for the provider who intubates a handful of times a year: a co-pilot that labels the glottis when the view is ugly and the stress is high, on top of a VL platform that already beats direct laryngoscopy on first-pass success. Until there is prehospital data, the fundamentals don't move: optimize the view, minimize desaturation, and confirm every tube with waveform capnography.

Sources

• NEJM (PMC) — DEVICE: video vs direct laryngoscopy in critically ill adults
• Resuscitation (PubMed) — Video laryngoscopy for out-of-hospital cardiac arrest, ESO national cohort
• BMC Anesthesiology (PMC) — AI-based airway identification system in tracheal intubation
• Anaesthesia (PMC) — larynGuide AI-assisted tracheal intubation in humans: diagnostic accuracy
• J Clin Monit Comput (PMC) — AI-assisted videolaryngoscopy: randomized simulation trial
• Italian Journal of Pediatrics (PMC) — Sensorized videolaryngoscope with force feedback for intubation training