Swallowing detection by sonic and subsonic frequencies: a comparison

Area of Science:

• Biomedical Engineering
• Acoustics
• Speech and Hearing Science

Background:

• Acoustic detection of swallowing is vital for diagnosing swallowing disorders and monitoring ingestive behavior.
• Swallowing sound detection can be compromised by artifacts from speech, chewing, and ambient noise.
• Investigating frequency ranges for improved acoustic swallowing detection is necessary.

Purpose of the Study:

• To compare the accuracy of swallowing detection using acoustic signals in subsonic (≤ 5 Hz) versus sonic (20-2500 Hz) frequency ranges.
• To determine if subsonic frequencies are less susceptible to artifacts from intrinsic sounds like speech and chewing.

Main Methods:

• Employed a simple method using throat and ambient microphones to capture acoustic signals.
• Acquired and analyzed swallowing sounds in both subsonic and sonic frequency ranges.
• Calculated and compared averaged recall and precision values for each frequency range.

Main Results:

• Averaged recall values exceeded 85% for both subsonic and sonic frequency ranges.
• Averaged precision values were 50% for subsonic and 42% for sonic frequencies, with high false positive rates in both.
• No significant difference in averaged precision values was observed between the two frequency ranges.

Conclusions:

• Subsonic frequencies did not demonstrate a significant advantage over sonic frequencies in reducing intrinsic sound artifacts for swallowing detection.
• The current acoustic detection method using these frequency ranges shows similar susceptibility to artifacts.
• Future research should incorporate a signal classification layer to potentially improve artifact differentiation and detection accuracy.