RF Sensing Based Breathing Patterns Detection Leveraging USRP Devices.

Area of Science:

• Medical Technology
• Artificial Intelligence in Healthcare
• Respiratory Monitoring

Background:

• Non-contact breathing pattern detection is crucial for remote healthcare and disease diagnosis, especially for COVID-19.
• COVID-19 is associated with distinct abnormal breathing rates, necessitating real-time monitoring for prediction and screening.
• Non-contact methods reduce patient burden, minimize infection risk, and lower environmental constraints during pandemics.

Purpose of the Study:

• To develop a platform for unobtrusive, remote detection and classification of abnormal breathing patterns.
• To leverage software-defined radio (SDR) and radio frequency (RF) sensing with machine learning (ML) for breathing analysis.
• To evaluate the performance of ML algorithms in accurately identifying diverse abnormal respiratory signatures.

Main Methods:

• Utilized software-defined radio (SDR) based radio frequency (RF) sensing techniques.
• Employed machine learning (ML) algorithms for the classification of detected breathing patterns.
• Evaluated ML model performance based on accuracy, prediction speed, and training time.

Main Results:

• The developed platform achieved a maximum accuracy of 99.4% in detecting and classifying abnormal breathing patterns.
• A complex tree algorithm demonstrated superior performance in the classification task.
• The system provides real-time detection and classification capabilities.

Conclusions:

• The SDR-based RF sensing and ML platform offers a highly accurate method for non-contact breathing pattern analysis.
• This technology has significant clinical implications for pandemic and non-pandemic respiratory health monitoring.
• The system can be deployed for practical applications in disease prediction, diagnosis, and screening.