To investigate the consistency between digital stethoscope technology and manual stethoscope in chronic obstructive pulmonary disease (COPD) patient monitoring scenarios.

This prospective diagnostic accuracy study employed non-random continuous sampling, collecting 1,953 valid lung sound recordings from COPD patients at the Department of Respiratory and Critical Care Medicine, Xinjiang Production and Construction Corps Hospital between November 2024 and May 2025. A multidimensional respiratory sound assessment scale was developed with standardized interpretation criteria. Clinical physicians evaluated digital lung sound characteristics while comparing artificial intelligence-powered digital auscultation results with physician assessments for consistency. Taking the clinical physician's listening results as the gold standard, the following indexes were calculated through the confusion matrix: accuracy, recall rate, false positive rate and precision, so as to evaluate the diagnostic consistency between digital stethoscope and manual stethoscope. This evaluation system assessed diagnostic consistency between digital stethoscope and manual stethoscope.

The accuracies of digital stethoscope and AI analysis in identifying normal lung sounds, coarse breath sounds, rales, dry rales, and wet rales were 99.39%, 99.59%, 99.18%, 99.00%, and 99.59% respectively; recall rates were 99.44%, 98.31%, 99.49%, 99.10%, and 98.71%; false positive rates were 0.760%, 0.125%, 1.025%, 1.059%, and 0.000%; precision rates were 99.11%, 99.43%, 98.48%, 96.49%, and 100.00%. Consistency tests showed that the Kappa values of digital stethoscope and AI analysis and clinicians' stethoscope judgment of normal lung sounds, coarse breath sounds, rales, dry rales and wet rales were 0.988, 0.984, 0.983, 0.972 and 0.992, respectively.

The digital stethoscope results showed a high degree of consistency with the manual stethoscope results in patients with COPD. Digital auscultation technology can alleviate the pressure on medical resources and reduce human errors through standardized analysis, and it is a key direction for the development of smart healthcare.