Subtle Sounds – Component Sound Analysis for Extracting and Analyzing Medical Information from Patient Encounters

Nearly 37 million people in the United States suffer from respiratory diseases such as asthma, chronic bronchitis and lung cancer. Physicians can’t capture all sound that occurs during doctors’ visits, so much of the nonverbal information from patient encounters is not extracted for further analysis.

The sound analysis system consists of three subsystems. A handheld device containing a digital stethoscope captures patient sounds using a Raspberry Pi Zero W, and a room audio capture system uses high-quality microphones to record the doctor-patient interaction. These two systems then transmit data to the third component: a server which analyzes and processes the recorded sound files using a Raspberry Pi4 and stores them securely within the clinic database. The system is equipped with various open-source codes using Python and MATLAB to extract and analyze sound components.