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

Project number: 
21049
Sponsor: 
UA Center for Accelerated BioMedical Innovation (ACABI)
Academic year: 
2020-2021
Project Goal: Design a system that detects heart, breath and speech through sound, which is analyzed using natural language processing technology to aid doctors in diagnosing patients.

Subtle Sounds is a holistic system that takes in arbitrary sound data and computationally analyzes the signals with a natural language processing, or NLP, interface.

An electronic stethoscope collects heart and breath sound signals, and it transmits data to a computer via Bluetooth. Four large diaphragm cardioid condenser microphones, mounted on individual tripods strategically positioned to surround the patient, capture speech sound signals. Parselmouth, a Python library for Praat software, determines the acoustic properties of input signals, including harmonic-to-noise ratio, jitter, shimmer, short-time energy, energy entropy and zero-crossing rate. The software also creates reports that depict relevant data and tables that are useful for a doctor's diagnosis. The software-generated reports are saved in a small on-board database.

Newly written software compatible with Google's API NLP provides transcription of the patient consultation. The software runs on a PC mounted on a mobile cart. The cart also holds the microphones, mounts, cables and stethoscope when the system is disassembled. A doctor or medical professional can easily transport the system and set it up in five minutes or less.

Get started and sponsor a project now!

UA engineering students are ready to take your project from concept to reality.