Generic implementation of a distress sound extraction system for elder care

Area of Science:

• Computer Science
• Biomedical Engineering
• Signal Processing

Background:

• Home-based remote patient monitoring offers enhanced comfort and cost savings.
• Existing systems utilize physiologic and position sensors, but lack auditory event analysis.
• Sound classification provides valuable supplementary data for patient monitoring.

Purpose of the Study:

• To implement a real-time smart sound system for detecting and identifying sound events in noisy home environments.
• To leverage generic PC hardware for cost-effective implementation of remote monitoring solutions.
• To enable rapid transmission of alarm information via the internet to monitoring centers or caregivers.

Main Methods:

• A real-time sound event detection module identifies potential alarm sounds.
• A parallel sound classification module segments events into sound or speech.
• For speech events, a speech recognition system is initiated; for sound events, classification into predefined categories is performed.

Main Results:

• The system successfully detects and classifies sound events in real-time, even in noisy conditions.
• Utilizes a standard PC with basic hardware (sound card, microphone) and an internet connection.
• Alarm notifications can be transmitted remotely via email or SMS.

Conclusions:

• The developed smart sound system offers a cost-effective and efficient method for enhancing home-based remote patient monitoring.
• Integration of sound analysis provides critical additional information beyond traditional sensors.
• The system facilitates timely alerts, improving patient safety and potentially reducing hospitalizations.