Irregular Breathing Classification from Multiple Patient Datasets - Prediction and Classification of Respiratory Motion - page 109

vector x 0 ¼ ð Þ . We denote the weight vectors w ji as the input-to-hidden layer
weights at the hidden neuron j and w kj as the hidden-to-output layer weights at the
output neuron k. Each output neuron k calculates the nonlinear function output of
its net activation U net k

ð Þ to give a unit for the pattern recognition.

6.3 Proposed Algorithms on Irregular Breathing Classifier

As shown in Fig. 6.1 , we first extract the breathing feature vector from the given
patient datasets in Sect. 6.3.1 . The extracted feature vector can be classified with
the respiratory pattern based on EM in Sect. 6.3.2 . Here, we assume that each class
describes a regular pattern. In Sect. 6.3.3 , we will calculate a reconstruction error
for each class using neural network. Finally, in Sect. 6.3.4 , we show how to detect
the irregular breathing pattern based on the reconstruction error.

6.3.1 Feature Extraction from Breathing Analysis

Feature extraction is a preprocessing step for classification by extracting the most
relevant data information from the raw data [ 18 ]. In this study, we extract the
breathing feature from patient breathing datasets for the classification of breathing
patterns. The typical vector-oriented feature extraction including Principal com-
ponent analysis (PCA) and Multiple linear regressions (MLR) have been widely
used [ 18 , 19 ]. Murphy et al. showed that autocorrelation coefficient and delay time
can represent breathing signal features [ 2 ]. Each breathing signal may be sinu-
soidal variables [ 4 ] so that each breathing pattern can have quantitative diversity of
acceleration, velocity, and standard deviation based on breathing signal amplitudes
[ 7 ]. Breathing frequency also represents breathing features [ 13 ].

Fig. 6.1 Irregular breathing
pattern detection with the
proposed algorithm

Feature
Datasets

Reconstruction
Error using NN

Probability (β 1 )
Class mean ( M 1 )
Covariance (Σ 1 )

Neural Network

th 1

Threshold
& Score

Σ

Class 1

1
δ

# Patient

Probability (β M )
Class mean ( M M )
Covariance (Σ M )

Neural Network

th M

Threshold
& Score

Class M

Σ

M
δ

# Patient