Best Apnea Interventions for Research (BestAIR)

• Best Apnea Interventions for Research (BestAIR) is a framework that combines CFD, LES, and LDA to evaluate obstructive sleep apnea treatments.
• It assesses surgical techniques like UPPP by reconstructing patient-specific airway models and quantifying outcomes using metrics such as AHI and area ratio.
• The approach links spectral flow analysis and recirculation cell characteristics to clinical success, guiding personalized surgical planning.

Best Apnea Interventions for Research (BestAIR) refers to the rigorous investigation and comparative evaluation of interventions for obstructive sleep apnea (OSA), focusing on quantifiable physiological, anatomical, and surgical outcomes. Advanced methodologies such as computational fluid dynamics (CFD), large eddy simulation (LES), and experimental flow validation are used to assess not only structural airway changes but also hemodynamic and spectral flow signatures linked to respiratory function. The paradigm has shifted from simple anatomical widening toward an integrative, flow-dynamics-guided framework to optimize OSA surgical planning and outcome prediction (Lu et al., 2020).

1. Surgical Intervention: Uvulopalatopharyngoplasty

The primary surgical intervention investigated in the context of BestAIR is uvulopalatopharyngoplasty (UPPP), a procedure targeting posterior-pharyngeal airway narrowing. UPPP involves resection of excess soft-palate tissue, uvula, and redundant lateral pharyngeal walls specifically in the retro-palatal region, aiming to enlarge the minimum cross-sectional area. Patient-specific pre-surgery and post-surgery airway morphologies are reconstructed from computed tomography (CT) scans (axial resolution 0.7 × 0.7 mm², 0.625 mm thick), and subsequently 3D-printed at anatomically realistic scale for in vitro experimentation. Clinical evaluation reveals that UPPP leads to variable outcomes: in a reported cohort, three patients demonstrated clinically significant reduction in apnea-hypopnea index (AHI), whereas one had worsened post-surgical AHI, underscoring that mere airway widening is not a reliable metric for functional success (Lu et al., 2020).

2. Experimental and Numerical Characterization

2.1 Laser Doppler Anemometry (LDA) Validation

Three-dimensional printed airway models are instrumented for laser Doppler anemometry (LDA) to validate numerical predictions of flow velocity. The inlet and outlet extensions minimize boundary layer artifacts. Axial velocity profiles are acquired at multiple cross-sections using a Dantec system (beam diameter 2.2 mm, ~2.5% uncertainty at 0.15 m/s), with synthetic oil fog (DEHS) as seeding particles and flow rates controlled to physiologic levels (16.8 L/min). Optical access is ensured by dedicated windows in the oropharynx; pre- and post-operative models are scanned on single or multiple radial lines per cross-section. LDA measurements confirm excellent correspondence with CFD velocities (including reversed flow), supporting the fidelity of the simulation methodology.

2.2 CFD and LES Methodology

CFD simulations leverage ANSYS Fluent 14.5 in an incompressible setting (Mach < 0.3), assuming rigid, no-slip walls and time-varying inspiratory profiles (12 breaths/min). Second-order space and time resolution, pressure–velocity coupling via SIMPLE, and a wall-adapting local eddy-viscosity (WALE) subgrid-scale model (Re ≈ 900–3 200) capture laminar-turbulent transitions. Meshes employ wall-normal refinement, grid convergence verified to Δ≈0.2%, and typical grids comprise 5–8 million cells. Pre- and postoperative models are compared to assess the effect of surgery on local flow field topology (Lu et al., 2020).

3. Spectral Flow Analysis and Signal-to-Noise Ratio

Spectral analysis of intrapharyngeal pressure time series, sampled at 128 Hz, reveals a distinctive 3–5 Hz oscillation attributable to flow separation and downstream recirculation. Morlet wavelet transforms $W(s,t)$ and fast Fourier transforms (FFT) are used to compute the power spectral density (PSD) $S(f)$, from which the amplitude of the 3–5 Hz signal ($A_s$) is measured against background noise ($A_0$).

The signal-to-noise ratio (SNR) at 3–5 Hz is defined as:

$$\mathrm{SNR}_{3–5\,\mathrm{Hz}} = 20\,\log_{10}\left(\frac{A_s}{A_0}\right)$$

This metric is inversely correlated with clinical severity, i.e., as AHI increases, $\mathrm{SNR}_{3–5\,\mathrm{Hz}}$ decreases. Quantitative analysis shows Pearson $r\approx-0.868$ ($p=0.005$) in simulated data and $r\approx-0.83$ in-vivo, establishing the utility of SNR as a physiomarker of post-intervention breathing quality (Lu et al., 2020).

4. Flow Separation, Recirculation, and Pressure Drop

Postoperative flow topology is characterized by examining recirculation zones and pressure gradients downstream of the retro-palatal constriction:

• Flow Separation Identification: LES velocity fields reveal regions of reversed axial velocity ($u_x<0$) immediately downstream of the constriction, visualized via streamlines/pathlines.
• Recirculation Strength: Quantified as $R = \int_{A_{rev}}|u_x|\,dA$ over the area of reverse flow.
• Pressure Drop: Instantaneous $\Delta P = P_{choanae} - P_{min}$ from choanae to minimum cross-section.

Clinically successful surgical outcomes consistently produce a single, strong recirculation cell with a robust 3–5 Hz SNR and moderate pressure drop. By contrast, failed cases display fragmented or multiple weaker recirculation zones with reduced spectral content and diminished beneficial oscillations (Lu et al., 2020).

5. Anatomical–Physiological Correlation: Area Ratio and AHI

A key anatomical metric is the area ratio (AR) of the minimum retro-palatal cross-section to the maximum cross-section behind the tongue base:

$$AR = \frac{A_{min,\text{retro-palate}}}{A_{max,\text{behind-tongue-base}}}$$

Empirical data demonstrate a strong negative correlation between $AR$ and AHI ($r=-0.833$, $p=0.01$, $N=8$ pre-/post-operative instances across four subjects). Lower $AR$ values—the result of a tighter retro-palatal narrowing—predict higher AHI. Post-operative increases in $AR$ are associated with successful outcome, supporting use of this geometric index as a preoperative planning metric (Lu et al., 2020).

6. Guidelines for Intervention Optimization and Clinical Integration

A synthesis of spectral, flow-dynamic, and geometric criteria provides a rational framework for surgical planning:

• Spectral Signal: A pronounced SNR at 3–5 Hz indicates likely success ($\mathrm{SNR}_{3–5\,\mathrm{Hz}}$ above a set threshold).
• Pressure Drop Moderation: Surgical widening should achieve a moderate reduction in pressure drop; overcorrection can eliminate beneficial flow oscillations.
• Flow Recirculation Structure: One stable recirculation cell is desired rather than multiple weak eddies.
• Anatomic Target: Post-operative $AR$ should exceed approximately 0.4–0.5.

Pre-operative planning integrates CT-based geometrical assessment and CFD. In cases where CFD predicts insufficient spectral or recirculation response, adjunctive procedures (e.g., tongue-base advancement) are considered. Post-operative validation may employ imaging and LDA/PIV to confirm restoration of the physiological 3–5 Hz oscillation. Combining these metrics enables prediction and verification of intervention efficacy beyond conventional anatomic targets, supporting the transition to a flow-dynamics-guided paradigm in OSA therapy (Lu et al., 2020).