Diaphragm physiological similarity index (DPSI): a de Novo speckle-tracking ultrasound metric for HFNC adjustment in acute respiratory failure.

BACKGROUND

Individualized diaphragm-protective respiratory support remains challenging, particularly during high-flow nasal cannula (HFNC) therapy. We developed the diaphragm physiological similarity index (DPSI) using speckle-tracking ultrasound to quantify diaphragmatic coordination and explore its potential to inform HFNC flow adjustment in acute respiratory failure (ARF).

METHODS

In this prospective observational study, 103 participants were enrolled: 45 healthy volunteers, 38 intensive care unit (ICU) patients for reproducibility assessment, and 20 ARF patients undergoing a prespecified stepwise HFNC flow adjustment protocol. Diaphragmatic displacement–time curves were acquired, and DPSI was derived from healthy reference waveforms. Reproducibility, feasibility, and physiological responsiveness to HFNC flow adjustments were evaluated and correlated with peak diaphragmatic excursion and the ROX (respiratory rate-oxygenation) index.

RESULTS

Healthy adults exhibited smooth, symmetric, and unimodal displacement-time curves, whereas ARF patients showed blunted, irregular, and variable trajectories. Mean inspiratory velocity (ICC = 0.87); peak displacement (ICC = 0.69); DPSI (ICC = 0.75) showed good reproducibility. DPSI increased during stepwise HFNC flow adjustments, with higher values commonly observed at intermediate flows (30–40 L/min) in many patients. DPSI was associated with the ROX index (R² = 0.62, p < 0.001) and peak diaphragmatic excursion (R² = 0.60, p < 0.001), reflecting physiological relevance.

CONCLUSION

DPSI is a feasible and reproducible speckle-tracking ultrasound metric that quantifies diaphragmatic motion patterns and distinguishes healthy from impaired contraction trajectories. DPSI responds to stepwise HFNC flow adjustments and may provide complementary physiological information to guide individualized HFNC management alongside established indices and clinical assessment. Future multicenter studies should evaluate its incremental value and association with patient-centered outcomes.

TRIAL REGISTRATION: ClinicalTrials.gov (NCT06996665, 3 May 2025).

SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s13054-026-05899-w.