Annemie Van Hirtum - annemie.vanhirtum@agr.kuleuven.ac.be
Aerts J.M., Moreaux B., Gustin P., Berckmans D.
Laboratory for Agricultural Buildings Research
Kardinaal Mercierlaan 92
B-3001 HEVERLEE, Belgium
Popular version of paper 3aABb6
Presented Wednesday morning, November 3, 1999
138th ASA Meeting, Columbus, Ohio
Modern society expends a lot of effort to optimize health care, in part by studying the evolution of infections in humans and animals. In this broad research area, observing symptoms and seeing how they evolve is obviously the key to a good diagnosis. This approach has resulted in the development and everyday use of several simple (e.g. thermometer) and complicated (e.g. NMR) medical tools.
Before making any diagnosis one must detect infection. Obviously, animals cannot verbalize a health problem. A person has to notice it for them. This might not cause a problem for controlling the health of a favorite pet, but for animals that stay in livestock buildings, monitoring their condition can be difficult because of the great number of animals in these buildings. Yet it is precisely the great density of animals in these spaces that makes it desirable to detect infections early. Automating the process of detecting animal infections would greatly improve efforts to monitor the health of these animals.
The principle proposed here involves interpreting animal sounds to to obtain signs of a particular infectious disease. In the case of respiratory infections, sound signals such as coughing and sneezing can hold relevant information usable as a detection system.
Before we are able to monitor animal sounds for signs of infection, we must identify the acoustical characteristics of sounds indicating infection. The research described here intends to characterize the sound of coughing, mainly by determining the pitches or frequencies of sounds associated with coughing. In other words, we are looking for "the music in coughing." The same way a musical note sounds different played on a piano or on a clarinet, it can be recognized as the same note to the experienced ear. The method employed in the coughing research so far is the same - unfortunately the sound not as melodious - in the sense that although each cough is not identical to the other depending on a person, animal, cause etc, it can by identified as coughing just by recognizing the key characteristics of the sound.