CFD in Sleep-Disordered Breathing Research: Applications and Insights

Disclaimer: The information in this post is for educational purposes only. It is not a medical diagnosis or treatment and should not be used to make healthcare decisions. Always consult a licensed provider for medical advice.

I.   Introduction

Sleep-disordered breathing (SDB) encompasses a spectrum of conditions including simple snoring, upper airway resistance syndrome (UARS), and obstructive sleep apnea (OSA). In all these conditions, airflow through the upper airway is compromised during sleep to varying degrees. The hallmark of OSA is repeated collapse of the throat (pharynx) due to complex interactions between airway anatomy and airflow mechanics. As a person with OSA inhales, a narrowed airway requires a greater negative pressure to draw in air; this increased suction further narrows the compliant soft tissues, creating a vicious cycle that can lead to airway collapse and noisy vibrations (snoring).

In UARS, the airway narrowing may not progress to a complete collapse, but even partial restriction causes significantly more negative intrathoracic pressure and increased breathing effort. These high negative pressures are thought to trigger brain arousals from sleep (despite the absence of full apneas) (1). In other words, UARS patients experience OSA-like fragmentation of sleep due to heightened breathing effort, even though standard sleep study metrics might classify them as “normal.”

Clinicians assess SDB severity with metrics like the apnea-hypopnea index (AHI) from sleep studies and examine anatomy via endoscopy or imaging. However, these approaches offer limited insight into the internal airflow dynamics that actually cause the airway to collapse or lead to arousals. This is where computational fluid dynamics (CFD) - a branch of engineering that simulates fluid (air) flow using the Navier–Stokes equations - has become a powerful research tool in SDB. Over the past two decades, CFD modeling has been applied to patient-specific airway anatomies (typically derived from MRI or CT scans) to virtually “peek” inside the airway during breathing and quantify parameters like airflow speed, pressure distribution, and resistance that are impossible to measure directly in a sleeping patient.

These models can reveal, for example, exactly where pressures drop along the airway during inspiration or how fast the air travels through a narrowed segment, helping to pinpoint the physical forces that lead to collapse. Below, we summarize how CFD has been used in SDB research, what insights it has provided to clinicians and scientists, and where this technology is heading in the future. We focus primarily on obstructive sleep apnea (the most studied SDB) while noting applications to related conditions (like UARS and primary snoring), highlighting use cases and peer-reviewed findings and avoiding unsupported treatment claims.

II.     CFD for Understanding Airflow Dynamics in OSA

Obstructive sleep apnea is fundamentally an airflow mechanics problem, so CFD is a natural fit for studying it. CFD modeling allows researchers to reconstruct a person’s upper airway in 3D and simulate inhalation or exhalation, revealing detailed flow patterns: where the air speeds up, where pressures drop, and how airflow interacts with the airway walls. These simulations have confirmed the mechanisms that cause airway collapse. (2) For example, as the throat’s cross-sectional area narrows (due to factors like obesity, enlarged tonsils, a relaxed tongue, etc.), the airflow must accelerate through the choke point – much like wind rushing through a tunnel. CFD quantifies this relationship: a smaller airway generates higher airflow velocity and more negative pressure (suction) on the walls. The elevated negative pressure can pull the compliant airway walls inward, contributing to collapse.  

Even if outright collapse does not occur, these heightened suction forces increase the work of breathing, which is precisely the issue in UARS, where the throat partially narrows and causes sleep disruption without full obstruction (8). CFD models also compute turbulence and shear forces; for instance, turbulent kinetic energy tends to spike in narrowed segments, correlating with the vibrations and noise of snoring. In essence, CFD gives us a visual and quantitative map of the breathing process in OSA that was previously hidden. One can pinpoint that a particular patient’s soft palate region experiences a large pressure drop and high airflow speed during inspiration, indicating a critical site of obstruction that is likely to collapse. (9),(10),(11)

CFD studies in OSA often use static anatomical models, which are essentially snapshots of the airway from a single phase of breathing (often from an awake scan or a breath-hold). These models are used to characterize normal vs. abnormal airflow patterns in the nose and throat, calculate airway resistance before and after surgeries, and even estimate the energy loss due to airway narrowings (12). Notably, CFD-derived pressure drop and resistance measurements often correlate with how symptomatic a patient is, more so than basic anatomical sizes alone. For example, one group found that pharyngeal airway resistance and pressure drop (CFD-derived metrics) in pediatric OSA patients correlated significantly with their apnea–hypopnea index, whereas the minimum anatomical cross-sectional area had a weaker correlation (13).

In that 2014 study, the authors concluded that “CFD model endpoints based on pressure drops in the pharynx were more closely associated with the presence and severity of OSAS than anatomical endpoints”, underscoring that airflow dynamics matter as much as anatomy (14). Another pediatric study similarly showed that the peak airflow velocity and the magnitude of airway negative pressure simulated in a child’s airway were strongly correlated with OSA severity (Spearman r ≈ 0.7, p < 0.01) (15). In that report, nearly all children with OSA had a very high maximum inspiratory airflow speed (often >12 m/s) compared to controls, and the regions of most negative pressure and fastest flow identified by CFD corresponded to each child’s obstruction site. In other words, two patients might have a similar-looking airway on MRI, but CFD can reveal which one experiences a worse pressure drop and turbulent airflow during inspiration – and that patient is likely to have more severe OSA.

It must be noted that airflow in the upper airway is highly dynamic – the airway caliber changes within each breath and across sleep stages. Most CFD simulations to date assume rigid airway walls (no movement) as a simplification. Recent research is now tackling the challenge of dynamic CFD modeling, where the airway shape moves or collapses during the simulation. For example, a 2024 study combined “cine” MRI imaging with CFD to add moving airway walls, and found that including the actual airway motion can significantly alter the airflow predictions (12). This line of work highlights a future direction: making CFD models more physiologically realistic by incorporating tissue elasticity, muscle forces, and time-varying deformation. While still largely in the research realm, these advances aim to bring CFD predictions even closer to what actually happens during a live sleep apnea episode.

CFD methods for the upper airway have been repeatedly validated against both invasive and laboratory measurements. In vivo studies using catheter-based pressure recordings have shown good agreement between CFD-predicted and measured pharyngeal pressures in OSA patients. (3) Complementary in vitro work has compared CFD to experimental flow data in anatomically accurate 3D-printed airway replicas, demonstrating strong correlations between predicted and measured pressure drops and resistances in both nasal and pharyngeal models (4) (5) (6). More recently, compliant pediatric airway phantoms have been used to show that CFD-derived pressures closely track applied external loads, further supporting physiological relevance. (7) Together, these efforts confirm that CFD can reliably reproduce measured airway pressures and resistances, strengthening confidence in its translational use for sleep-disordered breathing.

III.    Key Applications of CFD in Sleep-Disordered Breathing

Quantifying disease severity and risk: CFD can help estimate how “hard” it is for a patient to breathe through a given airway anatomy by computing pressure drops and flow resistances, and these metrics often correlate with or even predict clinical severity. As noted, CFD-derived pharyngeal pressure drops and resistance values have distinguished OSA patients from healthy controls and correlated with OSA severity better than anatomy alone (13),(15).

In one adult study, researchers performed CFD on awake CT scans and found that if the airflow speed in the narrowest airway segment exceeded roughly 3 m/s (or if the simulated pressure drop across that segment exceeded ~10 Pascals), the patient was very likely to have moderate-to-severe OSA (16). Using these CFD-derived thresholds, they could predict significant OSA with around 80% accuracy (area under the ROC curve ~0.78 for pressure drop). In that 2022 study, nearly 85% of patients with moderate/severe OSA had CFD-predicted airflow velocities ≥3 m/s and pressure drops ≥10 Pa, whereas those below these values were usually mild cases; indeed, higher CFD pressures/velocities were associated with worse overnight oxygen desaturation (16). Such findings suggest that CFD might be used as a screening or risk stratification tool – for example, a quick CFD analysis of anatomic scans could flag high-risk airway dynamics in patients who might otherwise wait months for a sleep study.

Identifying obstruction sites: CFD flow visualizations help pinpoint where the airway is most compromised. Often this isn’t obvious just from static anatomy – the narrowest section is not always the main problem. CFD simulations can reveal the regions of greatest pressure drop or highest velocity, indicating the functional “choke point” of the airway. In a pediatric OSA study, the location of maximum airflow velocity in the CFD model corresponded to the primary site of obstruction; notably, this was not always the same as the absolute smallest anatomical cross-section (15). In practical terms, CFD can detect if the true resistance “hotspot” is in the retropalatal area (behind the soft palate) versus the retroglossal area (near the tongue base), even if both regions look narrow on a CT scan. Knowing the primary obstruction site is crucial for tailoring treatments – it could determine whether a patient would benefit more from a tonsillectomy, a tongue-base reduction, or jaw advancement surgery. Some researchers have combined CFD with drug-induced sleep endoscopy (DISE) findings to validate these obstruction sites, while others use purely image-based CFD to non-invasively infer where collapse initiates. The takeaway is that CFD offers a kind of “functional endoscopy,” highlighting the airway zones that contribute most to airflow impairment in each individual (17),(15),(9),(18).

Virtual surgical planning and outcome prediction: One of the most impactful uses of CFD in SDB has been to evaluate and even forecast the effects of interventions. By modifying the airway geometry in the computer (for example, enlarging the jaw or virtually removing tissue) and re-running the airflow simulation, one can simulate a surgery or device and see if it actually improves airflow dynamics. There is a growing body of work doing this in retrospective and prospective studies.

For example, maxillomandibular advancement (MMA) – moving the upper and lower jaws forward – is a known cure for many OSA patients. A 2024 study took CT scans of OSA patients pre- and post-MMA and performed CFD on both. The surgery enlarged the upper airway volume by ~44% on average, and CFD showed a corresponding ~40% reduction in the peak airflow velocity (i.e. the air flowed more freely after skeletal expansion) (19).

The improvement in CFD metrics translated to clinical benefit: patients’ AHI (apnea–hypopnea index) dropped significantly as their airway volume went up and flow speed went down, with strong correlations between these changes. Interestingly, the authors noted that patients whose pre-surgery CFD showed extremely high airflow speeds tended to have better surgical outcomes than those with only moderately high speeds – and patients whose pre-surgery peak velocity was below ~7.2 m/s had a much higher chance of surgery “failure” (continued OSA despite MMA) (19). This hints that CFD might be used before surgery to identify good candidates (those who truly have a flow-limited airway that surgery can fix). In this study, a preoperative maximum velocity <7.2 m/s was associated with a 43% surgical failure rate, whereas those with more severe preoperative flow limitations almost universally improved. Surgeons could thus avoid invasive jaw surgery in patients whose CFD model predicts minimal benefit.

Evaluating less invasive treatments (devices and therapies): Beyond surgery, CFD has been applied to test oral appliances and other non-surgical interventions. In a 2019 study, CT scans of OSA patients before vs. after using a mandibular advancement splint (a type of oral appliance) were converted to CFD models (20). With the jaw advanced by the device, airflow velocity through the throat dropped significantly, and the overall airway resistance fell by about 36% (for example, from ~290 to 186 Pa/L in one group’s analysis). The simulations also showed that the negative pressure pulling on the soft tissues was reduced when the appliance was in place. These CFD results dovetail with the clinical goal of oral appliances: by lessening airway suction pressures and resistance, the device makes the airway less prone to collapse.

Similarly, CFD has been used to examine how continuous positive airway pressure (CPAP) works from a flow perspective (21). CPAP is essentially a steady positive air pressure applied to the airway via a mask, and CFD confirms that raising the inlet pressure (e.g. to 10 cm H₂O, a typical therapeutic level) effectively splints open the airway and increases static pressure throughout, without causing harmful high-velocity jets or excessive shear stresses on the airway walls. This provides reassurance that CPAP’s benefits (preventing collapse by pressurizing the airway) are not offset by any damaging flow phenomena – in fact, one CFD study found that under CPAP, airflow velocities did not increase to problematic levels and wall shear stress in the throat actually decreased despite the higher pressure.

On another front, researchers have evaluated nasal interventions with CFD, since nasal obstruction can worsen snoring and OSA. A recent CFD analysis of a nasal dilator device (which props open the nasal valves) showed about a 24% improvement in nasal airflow patency on the narrowed side of a deviated septum, but also found that the dilator’s presence slightly increased airflow resistance on the opposite side (22). The simulation output included detailed airflow, heat, and humidity profiles in the nasal passages, providing objective data on a device whose clinical benefits have been somewhat mixed in subjective reports. This ability to objectively measure the impact of an intervention – be it a surgery, oral appliance, CPAP, or nasal aid – in a patient-specific model is a key strength of CFD in SDB research.

Mechanistic insights (snoring and flow patterns): CFD and related fluid–structure simulations have also shed light on the mechanics of snoring and airway collapse. For example, a fluid–structure interaction (FSI) study in 2017 created a coupled CFD and finite-element model of the upper airway’s soft tissues to investigate snoring vibrations (23). The researchers found that the tongue base and soft palate in their model have natural vibration modes in the low-frequency range (approximately 12–40 Hz), and these structures’ oscillation frequencies correspond to components of the snoring sound . In the simulation, modes at ~21 Hz and ~39 Hz involved the soft palate vibrating, while other modes (e.g. ~12 Hz, 18 Hz) were driven by the tongue base – consistent with the idea that both regions can contribute to snore noise. The model was even able to reproduce how the soft palate and tongue might flutter as air passes through, with the CFD component showing the pressure and flow patterns that drive those vibrations. One insight was that very low-frequency oscillations (<20 Hz), which are difficult to discern in a snore audio recording, were present in the tissue vibration modes and therefore could be contributing to airway wall fatigue or disturbance.

This kind of research improves our understanding of why certain anatomical structures tend to vibrate and cause noise. It’s a step toward linking the subjective symptom (a patient’s snore sound) to objective airflow phenomena. Additionally, CFD studies often visualize flow patterns such as vortices forming in the pharynx or how airflow may shift between nasal vs. oral breathing routes; these visualizations help clinicians appreciate, for instance, why breathing through the mouth (bypassing nasal resistance) might worsen collapse at the throat. Overall, by mapping out airflow behavior, CFD has confirmed many aspects of OSA pathophysiology and provided new quantitative biomarkers (like pressure drop, flow velocity, wall shear stress) that complement traditional sleep-study metrics.

Detecting subtle flow limitations (UARS and mild SDB): CFD is not limited to frank obstructions; it can also identify and quantify more subtle airflow restrictions that may be clinically significant. UARS is a prime example: UARS patients experience OSA-like symptoms (daytime fatigue, fragmented sleep) due to high airway resistance and frequent arousals, yet they do not show the apneas/hypopneas that define OSA. Because their airway narrowing is milder, traditional diagnostic metrics (like AHI) often miss UARS. CFD, however, can illuminate the hidden airflow stress in these cases. A CFD simulation of a UARS patient’s airway during inspiration may reveal that even a moderate narrowing generates a disproportionately large pressure drop and high airflow velocity – essentially quantifying the extra effort needed to breathe. In principle, this provides concrete evidence of a problem that was previously only inferred via invasive esophageal pressure measurements (the gold standard for UARS detection (1).

In fact, researchers have noted that elevated airway resistance is associated with conditions like UARS, and that lowering resistance leads to better breathing and fewer arousals (8). This suggests that CFD might be used as a diagnostic or research tool for milder SDB – identifying patients who have pathological airflow dynamics (e.g. excessive negative pressure or turbulence) even when their anatomy appears “normal” and their sleep study is inconclusive (2).  By quantifying flow limitation in a personalized way, CFD could help bridge the gap between subjective symptoms and objective findings in mild SDB, offering a more nuanced assessment of who might benefit from intervention even if they don’t meet the formal OSA criteria. (It’s worth noting that this application is still in early stages; few peer-reviewed studies have focused specifically on UARS with CFD, but it remains an intriguing area for future research given CFD’s sensitivity to flow changes.)

IV.   Evolution and Future Directions

CFD in sleep medicine has evolved from a purely academic exercise to a potential clinical support tool. Early CFD models (2000s–2010s) established feasibility – showing that patient-specific upper airway flow could be simulated and correlated with clinical observations. In the current decade, we are seeing CFD used in more translational research and clinical pilot studies: for example, to predict who needs surgery, to test device efficacy, or to triage patients by risk. A notable trend is combining CFD with modern imaging and AI techniques to streamline the process. One group recently used a deep learning algorithm to automatically segment upper-airway CT scans, ran CFD on these models to generate flow features, and then trained a machine-learning classifier to diagnose OSA based on those features. Their automated approach achieved ~81% accuracy in distinguishing moderate OSA patients from healthy controls (with sensitivity ~89% and specificity ~86%) (24). The vision is that in the future, a clinician might obtain an anatomical scan and an AI-powered CFD pipeline could instantly output risk metrics (like a predicted pharyngeal pressure drop or airway resistance) and even suggest the most likely site of collapse or the best treatment option.

Another future direction is the integration of CFD with physiology – moving beyond airflow in a rigid tube to a more holistic model of the patient’s airway behavior. This could involve true fluid–structure interaction (actively coupling the airflow simulation with models of soft-tissue mechanics and muscle activation) as well as rigorous validation against real measurements. Progress is being made: dynamic CFD models using cine MRI are one example (as discussed above) (12).

A small but meaningful body of work shows that virtual or predictive CFD can forecast treatment success in OSA. In a seminal “virtual intervention” study, De Backer and colleagues digitally advanced the mandible in patient-specific upper-airway models and used CFD to predict reductions in pharyngeal resistance and pressure drop; those predictions aligned with the actual improvements measured after patients started mandibular advancement therapy, making it one of the earliest proofs of concept for CFD-based outcome prediction. (18) Building on that idea in a prospective clinical setting (though not strictly a “virtual surgery”), Van Gaver et al. used functional imaging with CFD before treatment to improve patient selection for mandibular advancement devices, demonstrating that pre-treatment CFD features prospectively predicted therapy outcome. (25) For skeletal surgery planning, several groups have explored preoperative/virtual geometry changes to guide maxillomandibular advancement (MMA)—for example, simulating directional/distance movements to anticipate aerodynamic benefit—indicating a path toward CFD-assisted surgical planning and prediction in bony procedures. (26)

These studies highlight how CFD can move beyond retrospective correlation into the realm of true predictive modeling, where virtual interventions provide a preview of whether a given surgery is likely to succeed for an individual patient. If these approaches are refined and validated in larger trials, clinicians could use CFD to plan surgeries with more confidence (choosing the right procedure for the right patient) and even to design personalized therapies (like a custom-fitted oral appliance optimized to a patient’s unique airway dynamics).

V.   Conclusion

Computational fluid dynamics has rapidly advanced our understanding of the hidden airflow characteristics in sleep-disordered breathing. It provides a non-invasive window into a patient’s airway, quantifying factors like pressure, velocity, and turbulence that directly contribute to obstruction and symptoms, whether those symptoms manifest as outright apneas (OSA) or as more subtle arousals due to flow limitation (UARS and heavy snoring). From characterizing how and where the airway collapses, to predicting disease severity, to virtually testing interventions, CFD has proven to be an invaluable research tool. The insights gained - many backed by peer-reviewed studies - are guiding a more personalized approach to managing OSA and related conditions. While CFD is not yet a routine part of clinical care (owing to the complexity and time required for simulations), ongoing technological improvements and integration with AI are steadily breaking those barriers. We can envision a future where airflow modeling becomes a standard extension of imaging for SDB patients: helping sleep specialists and surgeons tailor treatments to each individual’s airway mechanics. The path from lab to clinic is still being charted, but the trajectory is clear – by marrying engineering with medicine, CFD is opening new frontiers in sleep apnea research, with the promise of improved patient outcomes grounded in a deep, physics-based understanding of how we breathe during sleep.

  References

1.          Exar, E. N., & Collop, N. A. (1999). The upper airway resistance syndrome. Chest, 115(4), 1127–1139.

2.          Faizal, W. M., Khor, C. Y., Shahrin, S., Hazwan, M. H. M., Ahmad, M., Misbah, M. N., & Haidiezul, A. H. M. (2024). Computational Fluid Dynamics Analysis of Varied Cross-Sectional Areas in Sleep Apnea Individuals across Diverse Situations. Computation, 12(1), 16. https://doi.org/10.3390/computation12010016.

3.          Mylavarapu G, Subramaniam D, Jonnagiri R, Gutmark EJ, Fleck RJ, Amin RS, Mahmoud M, Ishman SL, Shott SR. Computational Modeling of Airway Obstruction in Sleep Apnea in Down Syndrome: A Feasibility Study. Otolaryngol Head Neck Surg. 2016 Jul;155(1):184-7. doi: 10.1177/0194599816639544. Epub 2016 Apr 5. PMID: 27048669; PMCID: PMC5886361.

4.          Li, C., Jiang, J., Dong, H., & Zhao, K. (2017). Computational modeling and validation of human nasal airflow under various breathing conditions. Journal of Biomechanics, 64, 59–68.

5.          Cherobin, G. B., Voegels, R. L., Pinna, F. R., et al. (2021). Correlation between CFD-simulated and experimental nasal resistance. American Journal of Rhinology & Allergy, 35(1), 33–40.

6.          Van Strien, J., Shrestha, K., Gabriel, S., et al. (2021). Experimental validation of computational fluid dynamics simulations of nasal airflow. Physics of Fluids, 33(9), 091904.

7.          Ignatiuk, D., Li, C., McConnell, K., et al. (2022). Validation of computational fluid dynamics in a compliant pediatric upper airway phantom. Journal of Biomechanics, 139, 111093.

8.          Maggard, M. D., Sankari, A., & Cascella, M. (2023). Upper airway resistance syndrome. In StatPearls. Treasure Island (FL): StatPearls Publishing.

9.          Taherian, S., et al. (2019). Evaluation of human obstructive sleep apnea using computational fluid dynamics: A systematic review. Computer Methods and Programs in Biomedicine, 182, 105042.

10.        Na, J. S., Jung, H. D., Cho, H. J., Choi, Y. J., Lee, J. S., & others. (2019). Computational analysis of airflow dynamics for predicting collapsible sites in the upper airways: A preliminary study. Journal of Applied Physiology, 126(2), 330-340.

11.        Jeong, S.-J., Kim, W.-S., & Sung, S.-J. (2007). Numerical investigation on the flow characteristics and aerodynamic force of the upper airway of patient with obstructive sleep apnea using computational fluid dynamics. Medical Engineering & Physics, 29(6), 637-651.

12.        Xiao, Q., Gunatilaka, C., McConnell, K., et al. (2024). The effect of including dynamic imaging derived airway wall motion in CFD simulations of respiratory airflow in patients with OSA. Scientific Reports, 14, Article 17242.

13.        Zheng, Z., Liu, H., Xu, Q. et al. Computational fluid dynamics simulation of the upper airway response to large incisor retraction in adult class I bimaxillary protrusion patients. Sci Rep 7, 45706 (2017). https://doi.org/10.1038/srep45706.

14.        Wootton, D. M., Luo, H., Persak, J. A., et al. (2014). Computational fluid dynamics endpoints to characterize obstructive sleep apnea syndrome in children. Journal of Applied Physiology, 116(1), 104–112.

15.        Yanagisawa-Minami, A., Sugiyama, T., Iwasaki, T., & Yamasaki, Y. (2020). Primary site identification in children with obstructive sleep apnea by computational fluid dynamics analysis of the upper airway. Journal of Clinical Sleep Medicine, 16(3), 431–439.

16.        Chung, W.-S., & Chung, S. (2022). Prediction of moderate to severe obstructive sleep apnea using neck computed tomography with computational fluid dynamics study. Frontiers in Medicine (Lausanne), 9, 838367.

17.        Marques, M., et al. (2018). Retropalatal and retroglossal airway compliance in patients with obstructive sleep apnea. Respiratory Physiology & Neurobiology, 254, 31–36.

18.        De Backer, J. W., et al. (2007). Functional imaging using computational fluid dynamics to predict treatment success of mandibular advancement devices in sleep-disordered breathing. Journal of Biomechanics, 40(16), 3708–3714.

19.        Furundarena-Padrones, L., Cabriada-Nuño, V., Brunsó-Casellas, J., et al. (2024). Correlation between polysomnographic parameters and volumetric changes generated by maxillomandibular advancement surgery in patients with obstructive sleep apnea: a fluid dynamics study. Journal of Clinical Sleep Medicine, 20(3), 371–379.

20.        Song, B., Li, Y., Sun, J., et al. (2019). Computational fluid dynamics simulation of changes in the morphology and airflow dynamics of the upper airways in OSAHS patients after treatment with oral appliances. PLoS ONE, 14(11), e0219642.

21.        Saha, S. C., Huang, X., Francis, I., & Saha, G. (2024). Airway stability in sleep apnea: assessing continuous positive airway pressure efficiency. Respiratory Physiology & Neurobiology, 325, 104265.

22.        Lee, K. B., Ventosa-Molina, J., & Fröhlich, J. (2024). Computational analysis of nasal airflow and its alteration by a nasal dilator. Computers in Biology and Medicine, 178, 108634.

23.        Luo, H., Scholp, A., & Jiang, J. J. (2017). The finite element simulation of the upper airway of patients with moderate and severe obstructive sleep apnea–hypopnea syndrome. BioMed Research International, 2017, Article ID 7058519.

24.        Ryu, S., Kim, J. H., Yu, H., et al. (2021). Diagnosis of obstructive sleep apnea with prediction of flow characteristics according to airway morphology automatically extracted from medical images: CFD and artificial intelligence approach. Computer Methods and Programs in Biomedicine, 208, 106243.

25.        Van Gaver, H., Op de Beeck, S., Dieltjens, M., De Backer, J., Verbraecken, J., De Backer, W. A., Van de Heyning, P. H., Braem, M. J., & Vanderveken, O. M. (2022). Functional imaging improves patient selection for mandibular advancement device treatment outcome in sleep-disordered breathing: A prospective study. Journal of Clinical Sleep Medicine, 18(3), 739–750.

26.        Ogisawa, S., Shiba, T., Kamei, K., & Yagi, T. (2019). Computational fluid dynamics analysis for the preoperative evaluation of maxillomandibular movement in obstructive sleep apnea. Journal of Cranio-Maxillofacial Surgery, 47(10), 1626–1632.