Temporal dissociation of physiological adaptations during speaking valve training in tracheostomized patients: a wearable device-based study

This is the authors' abstract. We don't add key points for this paper.

Background Speaking valve (SV) training facilitates communication and ventilator weaning for tracheostomized patients, yet the dynamic, multi-system physiological responses and their temporal trajectories remain poorly characterized. This study aims to comprehensively evaluate the cardiopulmonary adaptations during and following SV training using wearable sensing technology. Methods A prospective observational study was conducted on 120 tracheostomized patients. Continuous physiological monitoring was performed using a wearable sensor across three phases: baseline (T0), SV intervention (T1), and recovery (T2). A multidimensional panel of 86 parameters was extracted across four domains: macro vital signs, respiratory cycle structure, respiratory complexity, and heart rate variability (HRV). Results SV training was well-tolerated, with no participants meeting predefined safety stopping criteria. Analysis revealed a distinct temporal decoupling in physiological responses—that is, respiratory mechanical and autonomic adaptations occurred immediately during valve use, whereas oxygenation improvements emerged only after valve removal. During the intervention (T1 vs. T0), patients exhibited immediate respiratory mechanical adaptation and significantly enhanced respiratory complexity (increased multiscale entropy at scales1–5; adjusted p < 0.05), reflecting active neural compensatory regulation by the respiratory controller. This was accompanied by acute cardiovascular stress, evidenced by vagal withdrawal (decreased RMSSD, HF power, and total power). However, oxygenation metrics remained unchanged during T1. Notably, significant improvements in oxygenation (increased minimum SpO₂ and reduced desaturation burden) emerged exclusively during the recovery phase (T2 vs. T0; adjusted p < 0.05). Furthermore, autonomic recovery was delayed: the LF/HF ratio rose significantly from training to recovery ( p = 0.02), indicating persistent sympathetic predominance after valve removal. Conclusions SV training triggers an immediate respiratory regulation response and acute cardiovascular stress, whereas the beneficial effects on oxygenation manifest as a delayed phenomenon during recovery. This temporal decoupling underscores the necessity of incorporating monitored recovery periods into SV clinical protocols and highlights the value of multidimensional wearable monitoring in revealing covert physiological dynamics in critically ill patients.