Low-power wearable respiratory sound sensing.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Wearable Technology

Background:

• Automated recognition of respiratory sound patterns is crucial for continuous health monitoring.
• Asthmatic wheezing detection requires efficient algorithms for wearable devices.
• Low-power consumption is essential for long-term autonomy of wireless sensors.

Purpose of the Study:

• To propose a wearable wireless sensor for on-board respiratory sound acquisition and classification.
• To optimize the power consumption of the digital signal processor (DSP) for continuous monitoring.
• To evaluate asthmatic wheeze detection algorithms for low-power implementation.

Main Methods:

• Implemented and analyzed four asthmatic wheeze detection algorithms on a low-power TMS320C5505 DSP.
• Evaluated algorithm classification accuracy using a dataset of prerecorded respiratory sounds.
• Measured algorithm execution times and processing load on the DSP.

Main Results:

• Achieved a classification accuracy of over 92% with the best-performing algorithm.
• The optimal algorithm occupied only 2.6% of the DSP's processing time.
• Identified an algorithm featuring time-frequency tracking of wheezing crests with spectral energy features as the most efficient.

Conclusions:

• A wearable wireless sensor with an optimized DSP can effectively monitor asthmatic wheezing.
• Low-power DSP processing is key to achieving long autonomy for continuous respiratory monitoring.
• The developed system enables efficient and accurate detection of asthmatic wheezing symptoms.