Videolaryngoscopy is increasingly important in airway management, but optimal training methods remain unclear. Artificial intelligence may enhance procedural skill acquisition through real-time feedback and guidance. This high-fidelity simulation pilot study characterised the learning curve of artificial intelligence-assisted videolaryngoscopy using the LarynGuide™ system in a mixed cohort of medical students, anaesthesia residents and qualified anaesthetists with limited or no prior experience of videolaryngoscopy. Ninety-nine healthcare practitioners (medical students, medical doctors, anaesthesia residents and specialists) performed 10 intubation attempts on a high-fidelity simulator with artificial intelligence-assisted real-time feedback. Optimal intubation was defined a priori as ≥90% probability of time to intubation <30 s combined with a percentage of glottic opening >90%. Tracheal intubation success, defined as completion of tracheal tube placement, irrespective of the number of laryngoscopy passes, was achieved in 951 of 990 attempts (96%), with mean (SD) time to intubation of 16 (13) s. At the group level, basic procedural competency (≥90% cumulative success rate, assessed using cumulative sum analysis) was reached at the fourth attempt. Success rates plateaued thereafter, time to intubation continued to fall until the seventh attempt, and the percentage of glottic opening continued to improve through the tenth attempt. Exploratory modelling estimated that more than 10 attempts would be required to reach a 90% probability of optimal intubation. Anaesthesia residents performed better than participants with no prior intubation experience: higher success rate (98% versus 93%, p = 0.003), better glottic visualisation (mean percentage of glottic opening 90% versus 78%, p < 0.001) and shorter time to intubation (14 s versus 18 s, p = 0.021). Participants with prior video game experience reached optimal intubation thresholds four attempts earlier than non-gamers (p = 0.018). Outcomes were worse in the 60- to 70-year subgroup (n = 3, p < 0.001), while the participants' sex had no influence. Satisfaction ratings were high, with ease of use reported as 9/10. Artificial intelligence-assisted videolaryngoscopy facilitated rapid acquisition of basic competency within four attempts in this pilot simulation cohort. Exploratory modelling suggests differences from historical benchmarks which warrant further investigation.
Learning curves and demographic influence on skill acquisition of artificial intelligence‐assisted videolaryngoscopy training with the LarynGuide™ system: A pilot simulation study
Sorbello, M.Conceptualization
;Maniaci, A.;Pappalardo, F.;La Via, L.
2026-01-01
Abstract
Videolaryngoscopy is increasingly important in airway management, but optimal training methods remain unclear. Artificial intelligence may enhance procedural skill acquisition through real-time feedback and guidance. This high-fidelity simulation pilot study characterised the learning curve of artificial intelligence-assisted videolaryngoscopy using the LarynGuide™ system in a mixed cohort of medical students, anaesthesia residents and qualified anaesthetists with limited or no prior experience of videolaryngoscopy. Ninety-nine healthcare practitioners (medical students, medical doctors, anaesthesia residents and specialists) performed 10 intubation attempts on a high-fidelity simulator with artificial intelligence-assisted real-time feedback. Optimal intubation was defined a priori as ≥90% probability of time to intubation <30 s combined with a percentage of glottic opening >90%. Tracheal intubation success, defined as completion of tracheal tube placement, irrespective of the number of laryngoscopy passes, was achieved in 951 of 990 attempts (96%), with mean (SD) time to intubation of 16 (13) s. At the group level, basic procedural competency (≥90% cumulative success rate, assessed using cumulative sum analysis) was reached at the fourth attempt. Success rates plateaued thereafter, time to intubation continued to fall until the seventh attempt, and the percentage of glottic opening continued to improve through the tenth attempt. Exploratory modelling estimated that more than 10 attempts would be required to reach a 90% probability of optimal intubation. Anaesthesia residents performed better than participants with no prior intubation experience: higher success rate (98% versus 93%, p = 0.003), better glottic visualisation (mean percentage of glottic opening 90% versus 78%, p < 0.001) and shorter time to intubation (14 s versus 18 s, p = 0.021). Participants with prior video game experience reached optimal intubation thresholds four attempts earlier than non-gamers (p = 0.018). Outcomes were worse in the 60- to 70-year subgroup (n = 3, p < 0.001), while the participants' sex had no influence. Satisfaction ratings were high, with ease of use reported as 9/10. Artificial intelligence-assisted videolaryngoscopy facilitated rapid acquisition of basic competency within four attempts in this pilot simulation cohort. Exploratory modelling suggests differences from historical benchmarks which warrant further investigation.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
