Feedback enhances training and elicits biometric responses in novel Flexible Nasolaryngoscopy Model

Abstract

Objectives:

To demonstrate the feasibility and impact of auditory feedback in flexible fiberoptic nasolaryngoscopy (FNL) training with a novel model incorporating audio-triggers based on known anatomical loci of negative patient feedback

Methods:

This proof-of-concept study used a CT-based 3D-printed and silicone-casted model of nares to mainstem bronchi to teach FNL during an otolaryngology simulation training. The spatially-tracked laryngoscope tip triggered simulated responses (pain, gag, cough) when co-localized to specific anatomic landmarks: the nasal septum, epiglottis, and trachea. Participants wearing heart rate (HR) monitors completed three FNL trials, two without and one with auditory patient responses. Outcomes, stratified by prior trainee experience, included participant HR, survey-based changes in confidence, and assessment of auditory feedback utility.

Results:

Twenty-nine trainees ranging from medical student to fourth-year otolaryngology residents rated model realism and tactile experience 7.4±1.6 and 6.9±2.2 respectively on a ten-point Likert scale (mean±standard deviation). Average HR increased from resting baseline during all trials, decreased during the second trial, and increased in the final trial that included audio feedback (p=0.02). The least-experienced participants reported the greatest change in confidence (novice: +175%, low: +165%, intermediate: +73%, experienced: +13%). All trainees felt auditory feedback added value and improved fidelity.

Conclusion:

Heart rate increased from baseline during simulated FNL on a novel 3D-printed model, implying sympathetic activation and increased stress. HR decreased with repetition, implying improved comfort, but increased with simulated auditory patient responses, suggesting that patient feedback can increase trainee attentiveness. Future studies may analyze widespread implementation of this model for multi-disciplinary learners.

Graphical Abstract

Simulated auditory patient responses improve fidelity and add value during flexible nasolaryngoscopy training. Heart rate increased from baseline during simulated FNL on a novel 3D-printed model, implying sympathetic activation and increased trainee stress. HR decreased with task repetition, implying improved comfort, but increased with simulated auditory patient responses, suggesting that patient feedback can increase trainee attentiveness.

INTRODUCTION

Flexible nasolaryngoscopy (FNL) is among the most frequently performed procedures in otolaryngology and risks patient comfort, mucosal injury, and epistaxis. Unintentional stimulation during the insertion and advancement of the scope, particularly in sensitive patients (e.g., with irritable larynx syndrome or chronic cough), can not only cause discomfort but may confound exam findings by inciting hyper-functional activation. Successful execution requires technical mastery and responsiveness to visual and tactile feedback, as well as patient reactions. Despite the intricacy and risks, FNL training has historically been limited to learning on static models without accurate tactile or patient feedback, or learning at point of care, raising patient safety concerns.^1–2

Simulation technologies are widely used in medical education, offering controlled, repeatable procedural training without risk to patients; improved technical skills, medical decision-making, and communication among other outcomes, have widely been demonstrated^3–7. In otolaryngology, 3D printed models are increasingly used and have been shown to realistically replicate the human anatomy and improve procedural learning.^8–9 For FNL specifically, the use of high-fidelity models has been shown to improve performance when coupled with video feedback.¹⁰ The integration of virtual reality and/or other audiovisual guides further enhances the simulation experience, incorporating real-time and/or interactive elements that have been associated in other specialties with reduced procedure time, and increased patient comfort.^11–17

In the clinical setting, patient responses offer additional feedback that drives operator technique and yet is not routinely incorporated in simulation task training. Despite substantial advances in simulation technologies, capturing the complexity of real-world situations – where patient responses can vary based on anatomy, pathology, emotional state, and physiological reactions – remains a challenge. Further, the effect of patient responses on trainee stress, confidence, and performance remains incompletely understood.

This proof-of-concept study sought to assess the utility of a novel CT-based 3D-printed and silicone-casted FNL model which incorporated simulated auditory patient responses by examining their impact on biometric and psychological markers in trainees with varying degrees of prior experience.

MATERIALS AND METHODS

Study Design and Population

Adult (≥18 years of age) trainees of varied experience levels participating in an otolaryngology simulation training session at the University of Washington, were considered for inclusion. All study activities were reviewed and deemed exempt by the University of Washington Institutional Review Board (#51207).

Flexible Nasolaryngoscopy Model

A novel CT-based 3D-printed model of the head and neck with a silicone-casted airway from the nares to mainstem bronchi was designed for FNL simulation and training. The model was designed based on anonymized CT scans of an anatomically normal 17-year-old female. Facial features were smoothed and modified during the printing process to maintain patient anonymity. All aspects of the model were printed and cast true to the patient with exception of the pathway along the turbinates, which was modified to allow passage of a 4mm scope tip. To enhance haptic realism, materials of different physical properties were employed throughout the airway. The septum was 3D-printed in acrylonitrile styrene acrylate, a rigid plastic, while the turbinates and airway walls were cast from a 3D printed mold in pliable silicone (Dragon Skin 10, Smooth-On Inc.). An electromagnetic motion-tracking sensor (NDI TrackStar System, Model 90 Sensor, Waterloo, Ontario, Canada) was threaded through the working channel and positioned at the distal tip of the laryngoscope (Five-S 3.5 mm × 65 cm single use scope with 1.2 mm working channel, Storz, Tuttlingen, Germany). A second sensor was embedded in the posterior neck of the 3D-printed model. The laryngoscope’s tip location and orientation relative to the 3D-printed model was tracked in real time. Participants’ spatial trajectories were interpreted against an idealized FNL progress curve trajectory from nostril entry point (progress fraction 0.0) to carina (progress fraction 1.0) in 10% increments. Progress fractions were assigned according to the fractional progress associated with the closest point on the reference curve, allowing the plotting of each participant’s 3D trajectory onto an interpretable 2D curve, which also documented individual features such as pauses, backtracks, and rates of progress (Figure 1). Major transition points and anatomical landmarks were encountered at progress fractions between 30–40% (transitioning from nasopharynx to oropharynx), 45% (epiglottis/laryngeal inlet), and 60% (true vocal folds). If the laryngoscope tip contacted specific anatomical landmarks (nasal septum, skull base), simulated audible responses representing patient discomfort were triggered. Similarly, simulated gag and cough responses were prompted by contact with the epiglottis and trachea, respectively.

Figure 1.

Flexible nasolaryngoscopy trainer schematic (A) with overlaid airway lumen (B) annotated with incremental 10% progress fraction markers from the nostril entry (0.0) to carina (1.0); 3D-printed model with silicone casted airway (C)

FNL, Flexible Nasolaryngoscopy;

Data Collection

FNL Trials

Trainees completed three (3) FNL trials. After the first trial, trainees were given standard faculty feedback and guidance before proceeding with the second trial. The third trial incorporated simulated auditory patient responses, based on the laryngoscope tip location. For each trial, a digital trace from nares to carina was recorded in three dimensions within the model coordinate space, detailing laryngoscope direction and velocity, as well as instances where the user paused, reversed, or triggered auditory feedback. Biometric data was collected from participants during each trial via heart rate monitor.

Participant Experience and Model Feedback

Participants were administered surveys prior to and following completion of the three FNL trials. Participants reported prior FNL experience level, represented by the number of prior FNLs performed on human patients or in other simulated environments. Ten-point visual analog Likert scales were used to describe confidence with FNL before and after use of the model as well as to assess model realism, tactile experience, and perceived value of simulated auditory feedback.

Statistical Analysis

Participants were stratified into four categories, based on the number of previous FNLs performed on humans: novice (0), low (1 to 5), intermediate (6 to 15), and experienced (more than 15) (Table 1).

Table 1.

Experience designations based on number of prior flexible fiberoptic nasolaryngoscopy procedures performed

Experience Designation	Number of Prior FNLs Performed	Sample Size (n, %)
Novice	None	5 (17.2)
Low	1–5	6 (20.7)
Intermediate	6–15	4 (13.8)
Experienced	More than 15	14 (48.3)

FNL, Flexible Nasolaryngoscopy

Survey responses were analyzed in aggregate and compared using descriptive statistics and one-way Analysis of Variance (ANOVA) tests. Individual biometric data was aligned with corresponding FNL tracings and analyzed to identify differences within and between trials as well as anatomic and experiential factors associated with changes in heart rate.

RESULTS

Cohort Demographics

The analysis cohort consisted of 29 trainees ranging from medical student to PGY-4. Experience levels defined above correlated with the degree of training completed (e.g. PGY-year or student status), with medical students comprising the entirety of the novice group and senior residents (PGY 3–5) representing those most experienced.

Prior Experience and FNL Trajectory

More experienced trainees exhibited shorter FNL trial duration (seconds) than those with less prior experience (novice: 36.5±10.4; low: 34.6±12.7; intermediate: 18±3.9; experienced: 14.5±5.5) (p<0.001) (Figure 2).

Figure 2.

Mean FNL trial duration stratified by experience. Total time is averaged across trials for each participant and then averaged across participants within each FNL experience grouping. Error bars represent one standard deviation above and below the mean.

FNL, Flexible Nasolaryngoscopy;

The overall number of contacts between the scope and anatomical structures was higher among those with less prior experience (novice: 2.53±1.96; 1–5 low: 3.50±3.17) than those with more (intermediate: 0.83±1.11; experienced: 0.61±0.86) (p<0.001). Total time in seconds spent in contact with anatomical landmarks that triggered simulated audio feedback was also higher in the less experienced cohort (novice: 2.45±2.38; low: 4.29±5.69; intermediate:0.69±1.07; experienced: 0.41±1.01) (p<0.001) (Figure 3). This is illustrated in representative FNL tracings from individuals with both no and extensive prior experience (Figure 4).

Figure 3.

Undesirable anatomical contact during FNL. For each experience level, the average number of undesirable anatomical contact events (A) or total time spent in contact with undesirable anatomical regions (B). Data is averaged across trials for each participant and then averaged across participants within each FNL experience grouping.

FNL, Flexible Nasolaryngoscopy;

Figure 4.

Differences in FNL trajectory based on experience level: two representative examples shown, a novice (A) and an expert (B). Graphs show scope tip progress vs time with various quantified quality metrics marked. Identified pauses (horizontal lines above the progress curve), backtrack events (triangles), undesirable anatomical contact events (diamonds), and continuous contact with undesirable anatomical regions (thickened progress curve) are shown. In addition, events where auditory patient feedback would be generated are shown as shaded rectangles; these events are indicated in all three trials, for reference, but were only audible to participants in Trial 3 (also indicated by using darker shading). All the same quality metrics were assessed for both examples shown, but no pauses, backtracks, or undesirable anatomical contacts occurred for the expert FNL trajectory (B).

FNL, Flexible Nasolaryngoscopy;

Biometric Response to Feedback

Among all experience levels, heartrate trended downward by an average of 6 beats per minute between the initial trial (mean: 84.1, SD: 12.2), which was followed by faculty feedback, and the second trial (mean: 78.2, SD: 9.5). Heart rate increased by an average of 3 beats per minute in Trial 3 (mean: 81.4, SD: 11.0), with the integration of auditory simulated patient responses (Figure 5). While the group effect was not significant (p>0.05), this was impacted by small sample sizes in the less experienced trainee subgroups. Among the most experienced trainees, mean heart rate varied significantly between trials (p=0.02). Similar trends were seen in all groups except novice learners, which remained flat overall. Figure 6 illustrates the trend in heartrate for one representative experienced individual throughout and between each FNL trial.

Figure 5.

Differences in mean heart rate per FNL trial by experience level. Trial 1 represents the initial trial; trial 2 was performed following standard faculty feedback; trial 3 incorporated simulated auditory patient responses. Standard deviations across experience levels were as follows for trials 1, 2, and 3, respectively: Novice (8.3, 6.3, 7.9); Low (16.4, 14.6, 15.4); Intermediate (17.7, 11.4, 14.6); Experienced (6.6, 6.6, 6.8).

FNL, Flexible Nasolaryngoscopy;

Figure 6.

Heart rate over the course of FNL simulation participation for a representative experienced individual. Darker points are HR values recorded during an FNL trial. Data within each individual FNL trial is averaged (dashed lines).

FNL, Flexible Nasolaryngoscopy;

Model Utility and Realism

Participants of all experience levels reported increased confidence with FNL following use of the model (Table 2). Novice and low experience trainees reported greater average changes in confidence performing the procedure (+175% and +165% and respectively) than those more experienced (intermediate: +73%; experienced: +13%).

Table 2.

Trainee assessment of model utility stratified by prior FNL experience

Experience Level	Mean Percent Change in Confidence Level (%)	Model Utility (VAS 0–10) (mean, SD)
		Model Endoscopic Appearance Rating	Model Tactile Realism Rating	Trajectory Feedback Helpfulness Rating	Auditory Feedback Helpfulness Rating
Novice	175	7.8 (0.84)	7.6 (1.67)	9.25 (0.5)	8.6 (1.52)
Low	165	7.83 (1.94)	7.33 (1.63)	9.0 (1.26)	7.67 (2.94)
Intermediate	73	7.25 (0.5)	7.5 (0.58)	8.0 (0.82)	6.25 (1.5)
Experienced	13	7.14 (1.79)	6.29 (2.73)	7.88 (2.44)	7 (2.63)
P Value	p <0.001	p >0.05	p >0.05	p >0.05	p >0.05

FNL, Flexible Nasolaryngoscopy; VAS, Visual Analog Scale (rated 0–10)

On a ten-point Likert scale, participants rated the model’s endoscopic appearance 7.4±1.5 and tactile realism 6.9±2.2. Participants rated the helpfulness of trajectory and auditory feedback 8.3±1.9 and 7.3±2.4, respectively. Ratings of endoscopic appearance, tactile realism, and the helpfulness of both audio and trajectory feedback did not differ significantly between experience levels.

DISCUSSION

In this study, we found that simulated auditory patient feedback enhanced flexible fiberoptic nasolaryngoscopy training on a novel 3D-printed and silicone-casted model with integrated electromagnetic tracking technology, and elicited a biometric response among trainees with some prior FNL experience. While the second trial was associated with a decrease in these trainees’ mean heart rate, the addition of simulated auditory patient responses was associated with increased mean heart rate across all levels of trainee experience except novice learners. All trainees with less prior experience also exhibited greater time and numbers of contacts with undesirable anatomical structures than more experienced trainees, confirming construct validity of the model. The incorporation of simulated audio feedback was well-received and improved the trainee experience, suggesting it to be a valuable adjunct to standard simulation training.

Several prior studies have demonstrated the utility of high-fidelity simulation technology to improve procedural skill development, boost trainee confidence, and enhance skill retention^18–20. With the incorporation of auditory and trajectory information and feedback, the current model features eight of ten recommended conditions for high-fidelity medical simulations as described in a systematic review by Issenberg et al. These conditions include: provision of feedback, opportunity for repetitive practice, utility within an existing curriculum course, multimodality learning inputs for learners, utility in a controlled environment, active individual participation, defined trajectory goals, and construct validity²¹. The remaining elements, which include capture of clinical variation and offering several difficulty levels can easily be incorporated by 3D printing and silicone casting a range of anatomical configurations and pathologies.

Increases in heart rate are proposed to represent heightened attention and improved performance on cognitive tasks, especially those involving executive function^22–24. In the present study, heart rate decreased during the second trial, supporting associations between practice and increased comfort and familiarity in clinical and simulated settings^25–26. With auditory simulated patient responses, heart rates returned towards unpracticed levels, suggesting that this feedback elicited a stress response^27–28. Given prior studies of learner stress in educational settings showing that a moderate amount of stress enhances performance and learning,²⁹ features such as these may lead to faster performance gains, serving as cues prompting course correction for trainees. Unexpectedly, the most novice trainees did not demonstrate the biometric response pattern seen in participants with prior experience (Figure 5). This may reflect a greater awareness among participants with prior experience of the clinical relevance of this procedure and its potential to cause harm, highlighting the importance of understanding a trainee’s prior experience when designing task trainers.

This study is limited by its sample size and single-site design, which may affect the generalizability of the findings. Additionally, the self-reported nature of survey data presents risk for recall bias, which may affect experience level grading. Biases in recall of prior experience and skill development might be addressed in future study by incorporating post-hoc blinded expert review to evaluate trainee skill level and improvement between trials. Further investigation with increased sample size could yield insight into how trajectories are modified in response to audio feedback and determine which types of audio feedback are most salient and impactful. Future study also offers the potential to assess the longitudinal benefit and impact on learner retention of audio feedback in skill development and maintenance relative to other teaching methods, helping further establish the use of this technique in standard training curricula.

In practice, both the laryngoscope tip and shaft present sources for possible patient discomfort. In the present model, only contact with the laryngoscope tip is registered to inform the user how the scope is contacting sensitive structures. While anatomic cues in the current iteration of the 3D printed model were limited to common regions (e.g., septum, epiglottis), future adaptations could augment the complexity of the trajectory and integrate additional sources of patient discomfort. Varied clinical situations also present unique needs and influence examiner and patient experience. Future studies may explore different ergonomic scenarios, such as cross-body scoping, and variations in patients, monitor position, and training environment. Finally, further investigation of the utility of trainee biometric responses in simulation settings may allow educators to better use these metrics to assess educational interventions.

CONCLUSIONS

Our findings suggest that flexible fiberoptic nasolaryngoscopy training using a 3D-printed and silicone-casted model can be enhanced using anatomically-specific auditory feedback. Physiological responses were seen among all learners except novices, with derived tracking metrics allowing detailed evaluation of performance. These findings suggest that future FNL task trainer development should further integrate tracking data and physiological feedback to optimize endoscopy training.