Optimizing cervical breath sound acquisition protocol in dysphagia patients: an orthogonal experimental study

Area of Science:

• Biomedical Engineering
• Clinical Acoustics
• Speech-Language Pathology

Background:

• Dysphagia is common in stroke patients, necessitating reliable screening methods.
• Cervical auscultation is key for dysphagia detection, with AI-driven acoustic analysis offering a promising alternative.
• Current clinical practice lacks standardized protocols for acquiring cervical respiratory sounds.

Purpose of the Study:

• To determine the optimal acquisition protocol for cervical breath sounds in post-stroke dysphagia patients.
• To establish standardized procedures for acoustic analysis in dysphagia screening.

Main Methods:

• An L9(33) orthogonal experimental design was used, varying auscultation position, dressing material, and tape type.
• Cervical breath sounds were recorded from 298 dysphagia patients across nine protocol variations.
• Acoustic signal processing quantified signal-to-noise ratio (SNR), aggregate entropy, and average power.

Main Results:

• Auscultation position, dressing material, and tape type significantly impacted SNR, aggregate entropy, and average power (p < 0.05).
• The optimal configuration (A1B1C3) maximized SNR (37.35 dB) and average power (78.79 dB), while minimizing aggregate entropy (1.51 bits).

Conclusions:

• The optimal protocol involves direct skin contact on the lateral trachea below the cricoid cartilage, without dressings, secured by medical-grade tape.
• These findings provide evidence-based guidance for standardized clinical implementation of cervical auscultation in dysphagia assessment.