Characterizing the Learning Curve for Performing Surgical Airway Management 

Changyun Ma¹, Gabriel Gazetta¹, Paul Raj Reddipogu¹, Kaori Tanaka², Matthew Hackett³, Jack Norfleet³, Rahul⁴, Steven Schwaitzberg⁵, Suvranu De⁶, Lora Cavuoto^1* 

¹Department of Industrial and System Engineering, University at Buffalo, Buffalo, New York

²Department of Emergency Medicine, Jacobs School of Medicine and Biomedical Sciences, University at Buffalo, Buffalo, New York

³U.S. Army Futures Command, Combat Capabilities Development Command Soldier Center STTC, Orlando Florida

⁴Department of Biomedical Engineering, Rensselaer Polytechnic Institution, Troy, New York

⁵Department of Surgery, Jacobs School of Medicine and Biomedical Sciences, University at Buffalo, Buffalo, New York

⁶FAMU-FSU College of Engineering, Tallahassee, Florida

Introduction: This study investigated the learning curve of behavioral performance and associated physiological changes during cricothyrotomy (CCT) skill training. CCT is a lifesaving emergency airway procedure with a rare incidence, required in only 1% of critical airway cases in the emergency department.  This rarity poses significant challenges for training and retainingexpertise, as opportunities to observe and practice in real-world settings are limited, leading to the poor success rate of 66% in the US military and 75% in UK hospitals. ^[1-3] To address the limited evidence on CCT skill training, this study integrated multimodal measures to characterize the learning curve of skill acquisition and changes in psychophysiological activities across a longitudinal training process.

Methods: Ten medical students completed three days of CCT training, performing six repetitions per day on a SimMan 3G manikin. Each trial was limited to two minutes, and was separated by a two-minute rest period. Heart rate was recorded as inter-beat intervals (IBI) using a Polar H10 chest strap and gaze data was collected via Tobii Pro Glasses 2. Completion time across the 18 repetitions was modeled using Pegels’ discrete exponential learning curve model^[4]. Heart rate variability (HRV) and gaze metrics were analyzed for the first repetition and the last repetitions of each day using repeated-measures ANOVA. They were further compared, through a one sample t-test, with data of emergency medicine program residents from our prior study^[5] conducted under the same protocol.

Results: The statistical results are summarized in Figure 1.The learning curve of completion time over eighteen training trials showed a dynamic pattern in which completion time dropped rapidly and significantly (p < 0.05) during the first day, reached a level comparable to that of residents by the sixth repetition (p>0.05), and shifted to a more gradual rate of improvement thereafter. Additionally, a modest increase in completion time was observed at the onset of the second and third days, suggesting a potential forgetting effect between days. The exponential learning curve modeling, implemented within a nonlinear mixed-effects framework with random intercepts, provided a strong fit to the pooled trial-level data (R² = 0.83, RMSE = 6.35). Parameter estimates indicated a learning rate parameter at 0.68 (95% CI [0.64, 0.72]), with performance approaching an asymptotic completion time at 39.4 seconds (95% CI [38.01, 40.83]).

The HRV measures indicated that heart rate and the LF/HF ratio exhibited significant reductions by the twelfth (p=0.01) and eighteenth (p=0.04) repetitions, respectively, suggesting a shift toward parasympathetic predominance and improved stress adaptation. The significant decrease in the saccade-to-fixation count ratio indicated enhanced fixation stability and reduced scanning behavior. No significant differences were observed in pupil diameter or in the root mean square of successive differences (RMSSD).

Conclusion:  The learning curve of CCT technical skill acquisition showed a rapid and significant improvement on the first day, followed by a slower progression that eventually plateaued. In contrast, stress adaptation and attentional control, as reflected by HRV and gaze behavior, emerged from the outset but required a longer duration to achieve stable adaptation. These insights can inform the design of evidence-based training programs for emergency procedures, addressing the limitations of traditional behavioral metrics in capturing psychological and cognitive factors, and providing a more comprehensive perspective that ultimately enhances clinician preparedness and patient safety.

Reference: