Heartbeat pattern classification algorithm based on Gaussian mixture model

Nowadays, probabilistic neural networks have been frequently used to pattern discrimination in biological signals despite of non-stationary and individual characteristics of human subjects. In this study, a new approach was proposed to pattern classification for electrocardiography (ECG) signals based on Gaussian mixture model and logarithmic linearization. The objective of this study was to identify and classify QRS complexes on ECG patterns. For this purpose, a high performance method to classify and discriminate various ECG patterns was developed. Besides, a comparison algorithm which evaluates time series signals was established, and the limitation of its parameters was determined in order to attain high performance in ECG classification. The proposed algorithm has been tested on the data from 20 normal subjects and 22 additional normal data sets from MIT-DB database. After the improvement by the proposed algorithm, we observed 99.21% and 99.24% of recognition rates in ECG data from 20 normal subjects and MIT-DB database, respectively. The results showed that the proposed algorithm achieved a high performance to classify and discriminate various ECG signals.