Table 2 Accuracy of human auscultation

The accuracy of lung auscultation

<Correct detection rate by sounds>

[42]

 – Pulmonologists: 28% (abnormal bronchial sound), 90% (wheezes)

 – Pediatricians: 16% (abnormal bronchial sound), 83% (wheezes)

 – Interns: 13% (abnormal bronchial sound), 83% (wheezes)

Physicians’ classification of lung sounds from video recordings

<Multirater agreement (Fleiss’ κ) between observers>

[43]

 – Detailed categories: 0.04 (rhonchi), 0.43 (high-pitched wheezes)

 – Combined categories: 0.59 (wheezes), 0.62 (crackles)

Pulmonary auscultatory skills during training in internal medicine and family practice

<Identification rates by sounds>

[44]

 – Trainees of family practice: 0% (whispered pectoriloquy), 84% (expiratory wheeze)

 – Trainees of internal medicine: 1% (whispered pectoriloquy), 82% (expiratory wheeze)

 – Pulmonary fellows: 5% (whispered pectoriloquy), 100% (expiratory wheeze)

Comparing the auscultatory accuracy of health care professionals

<Correct detection rates by sounds>

[45]

 – Staff of internal medicine: 86.7% (wheezes), 96.7% (crackles)

 – Resident of internal medicine: 59.0% (crackles), 80.0% (wheezes)

 – Adult ICU nurses: 47.0% (crackles), 88.0% (wheezes)

The contribution of spectrogram for visualization sound in clinical practice

<Proper diagnosis rate of medical students>

[46]

 Normal sounds: 57% (sound) → 63% (plus spectrogram)

 Wheezes: 70% (sound) → 83% (plus spectrogram)

 Crackles: 53% (sound) → 70% (plus spectrogram)

 Stridor: 70% (sound) → 73% (plus spectrogram)