The airway hyperresponsiveness prediction value of digitalized lung sound collected during forced expiration in bronchial provocation test.

Background: This study investigated the acoustic characteristics of forced expiratory lung sounds during bronchial provocation tests and their predictive value for airway hyperresponsiveness (AHR).

Research design and methods: Participants underwent a bronchial provocation test with incremental methacholine doses (0.072-1.25 mg). Forced expiratory volume in the first second (FEV1) was measured using spirometry, with pre-saline FEV1 as the baseline. AHR was defined as a ≥ 20% decline in FEV1%Ref (FEV1 relative to baseline). Simultaneously, lung sounds were recorded from the right lower lung field. Thirty-five acoustic features were extracted from the first-second forced expiratory lung sound, including 24 spectral parameters, 5 mel-frequency cepstral coefficients (MFCCs), and 6 entropy parameters. Correlation analysis, group comparisons, and logistic regression were conducted to assess the relationship between acoustic features and AHR.

Results: Seventeen patients tested positive for AHR. AHR was associated with decreased spectral parameters (A1-A3, MFCC2-MFCC5) and increased spectral entropy (p < 0.05). Logistic regression identified effective power (PT) and MFCC5 as independent predictors, yielding an AUC of 0.856 (95% CI: 0.769-0.944).

Conclusions: Acoustic features of breath sounds can predict AHR, offering a potential noninvasive alternative to bronchial challenge tests.