Categorizing of inhaling behaviors using signal processing and data mining techniques

This is the authors' abstract. We don't add key points for this paper.

The study of respiratory forms a major application and publication in the medical field. It characterizes the abnormalities in the breathing pattern, which assists in selecting the appropriate treatment methods. In some cases, respiratory characteristics unravel and point out potential diseases. A medical team gathered data from randomly selected recruits. A huge dataset was prepared, which capture the volume and velocity of the breathed air from the target recruits. This paper presents the results of carrying out some of the signal processing, dimension reduction, and data mining techniques over this dataset. In particular, convolution filter, singular value decomposition and density-based spatial clustering of applications with noise were applied. Inhaling behaviors have been categorized into nine groups but with stochastic noise. Some of the groups are big enough and distinguishable to evaluate with the use of eight types of inhalers the model for velocity versus volume of inhaling. Results will be considered by the medical team for choosing the appropriate inhaler out of five types of inhalers appropriate for each group.