Quantification of ventilatory control in sleep apnea: from physiological insight to computable loop gain

Sleep-disordered breathing reflects the interplay of upper-airway collapsibility, sleep depth/arousability, and instability in ventilatory control. Ventilatory loop gain (LG) quantifies the latter as the ratio of ventilatory response to disturbance: values >1 indicate self-sustaining oscillations, whereas lower values denote stable control. Despite its clinical relevance, LG measurement remains largely confined to research laboratories because conventional protocols (e.g. controlled gas challenges, stepwise reductions in positive airway pressure) are invasive and technically demanding. Recent advances have enabled several computable LG estimation methods from signals available in routine polysomnography and selected home settings, creating a timely opportunity to translate LG beyond the laboratory. Indirect approaches include breath-hold maneuvers, cardiopulmonary-coupling metrics, respiratory self-similarity analysis, and data-driven or model-based estimation. Simplified surrogates improve accessibility but sacrifice physiological detail, whereas model-based methods (e.g. Phenotyping Using Polysomnography [PUP]) provide individualized LG profiles at higher requirements for signal quality and computation. Emerging evidence from model-based polysomnographic estimation indicates that a dynamic LG threshold near 0.7 may help identify patients who benefit from chemorespiratory stabilizers alongside obstruction-resolving therapies, whereas those with lower LG often respond adequately to anatomy-focused approaches alone; whether equivalent thresholds apply across estimation methods remains to be established. Prior reviews have emphasized physiology and phenotype-based care, but none have systematically compared LG assessment methods across a fidelity-feasibility spectrum or linked method choice to treatment selection and validation needs. This review synthesizes perturbation tests, signal-based surrogates, and model-based identification into a pragmatic framework with decision cues for screening versus confirmatory testing and priorities for clinical deployment.