Background

In today’s world and especially in the developing countries the ability to diagnose respiratory diseases in masses is limited by lack of time and precision to identify the problems. Also the human variable when listening for lung problems is an issue. Many other areas of medicine already benefit from systems that use algorithms to help human decision making. In the respiratory diseases this has yet to happen.

Technology Overview

PSU researchers have developed a deep-learning algorithm which uses a digital database of lung sounds to help identify diseases faster while becoming better at detecting them. The algorithm is self-learning and instead of using pre-labeling system that matches previous lung sounds and definitions the deep learning algorithm evaluates new sounds based on the previous sounds and learnings thereof.

Applications

The system has already been tested and used via mobile phone app and can be scaled the same way and also by using digital stethoscopes.

Before clinical use this brings great benefit for researchers who are tagging the lung sounds by listening to them bit by bit. When using this technology it becomes better with every new data point collected.

Also the rising number of digital stethoscopes eases the field of applications while direct mobile phone use can open up the opportunities especially in the developing countries.

Opportunity

Development opportunity with the right partner to plan larger data collection and validation. Further test and development into a product.

A Platform for Collecting Lung Sound Data Sets and its Application to Low-Cost Disease Diagnostics

IP Status

Seeking