A group of US researchers have invented a system which captures and analyses non-speech body sounds in order to identify potential health problems in advance.

The study of bodily functions can tell us a great deal about a person’s health. In the e-Health field, we recently reported on the SimSensei project, a system which studies a person’s psychological well-being by using a Kinect camera and provides a virtual psychologist. However, non-speech body sounds may also feed back key information on a person’s health. Researchers at Cornell University, the University of North Carolina and the University of Rochester in the United States have developed a mobile detection system which captures and automatically interprets body sounds in order to assess the person’s physiological condition and detect any looming problems.

Tailored sound-capture system

BodyBeat is a mobile sensing system that includes a piezoelectric* sensor-based microphone embedded in a 3D printed capsule filled with a soft silicone, which is placed on the person’s throat to capture subtle body sounds conducted through the body surface. This device is tailor-made, as existing microphones on the market are not designed to analyse sounds in depth other than those of the human voice. The microphone can detect subtle body vibrations and sounds generated by various actions such as pulse, breathing and eating.  The researchers tested the system on a panel of people then conducted an analysis of the acoustic characteristics of a range of actions – breathing, coughing, laughing, chewing, etc – and created a classification algorithm for non-speech body sounds. This data can be transmitted to an Android smartphone, where a dedicated app analyses the sounds to detect any dangerous signs.

Anticipating health problems

In their paper entitled ‘BodyBeat: A Mobile System for Sensing Non-Speech Body Sounds’, the researchers list two ways the system could be applied – food journaling and illness detection. BodyBeat has fundamentally been designed to anticipate medical problems. It can for example analyse the frequency and intensity of coughing or heavy breathing to help predict incipient pulmonary illness.  However, other commercial applications might include mobile apps able to for instance track food consumption on a daily basis. The existing PlateMate app is designed to track the calorific content of dishes eaten by the user, but it requires users to photograph all their meals. BodyBeat might perhaps be used to complement the PlateMate system by detecting when a person is eating or drinking. It could then send a reminder to take a photo on PlateMate, or be integrated into a system such as Google Glass. Going forward, the US academic researchers plan to work with doctors to detect and assess the significance of other non-speech body sounds such as sneezing, and differentiate conditions associated with specific types of coughing.

*piezoelectricity is electricity generated from pressure

By Eliane HONG