Automated Detection of Trigger Asynchronies During Initiation of Noninvasive Nocturnal Ventilation

Nocturnal noninvasive ventilation (NIV) can improve outcomes in severe COPD. However, optimal ventilator titration is challenging, and patient-ventilator asynchrony (PVA) is common, reducing comfort and perceived benefits. Optimal PVA detection in nocturnal NIV has yet to be optimized, with continuous manual review of ventilator tracings being practically unfeasible. This study compared how noninvasive surrogate measures of subject-initiated breathing activity, namely respiratory inductance plethysmography (RIP) and surface electromyography (sEMG), influence automated PVA detection in subjects with COPD receiving nocturnal NIV.

PVA, including ineffective effort, double-trigger, and auto-trigger, was identified using novel waveform-based detection algorithms applied to nocturnal NIV recordings from subjects initiated on nocturnal NIV at the University Medical Center Groningen, the Netherlands. Algorithms operated on 3 predefined waveform groups (1) ventilator pressure-flow alone; (2) pressure-flow + thoracic RIP; (3) pressure-flow + intercostal or diaphragm sEMG. Primary outcomes were the PVA (%) per asynchrony type. Comparisons between waveform groups were performed at the subject level.

Datasets from 14 subjects were suitable for analysis across all data groups. Median overall PVA (%, comprising ineffective effort, double-trigger, and auto-trigger) varied by data group 12.7% for pressure-flow alone, 13.7% for pressure-flow + thoracic RIP, and 22.7% for pressure-flow + intercostal or diaphragm sEMG. Differences between methods were observed, although interpretation is influenced by differences in analyzable breath sets across modalities.

Integrating noninvasive respiratory effort signals with pressure-flow analysis resulted in different estimates of PVA during nocturnal NIV initiation in subjects with COPD. Thoracic RIP and surface electromyography provided complementary information on respiratory effort beyond ventilator waveforms, but their utility was constrained by signal availability and quality. These findings reflect modality-dependent differences in estimated asynchrony burden and highlight both the potential and current technical limitations of multimodal monitoring during nocturnal NIV.