A decision support system for lung function diagnosis in infants

Abstract

This paper describes an automatic medical decision support system for lung function diagnosis in infants. The method is based on tidal breathing flow volume loops and air flow resistance measurements, obtained noninvasively from sedated infants. In order to obtain a diagnosis from these signals, a neural network classification algorithm has been implemented for pattern recognition purposes. At the moment, the system allows the classification of about 70% of the breaths, with a failure rate of 15% and 15% rejections.