A reliable feature selection algorithm for determining heartbeat case using weighted principal component analysis

This study proposes a simple and reliable feature selection algorithm for determining heartbeat case using Weighted Principal Component Analysis (WPCA) method. The proposed WPCA consists of four major processing stages. The first stage is the preprocessing stage for enlarging ECG signals' amplitude and eliminating noises. The second stage is the QRS extraction stage for detecting QRS waveform using the Difference Operation Method (DOM), which includes two processes, one is the difference operation process, and the other is the waves' detection process. The third stage is the qualitative features stage for feature selection on ECG signals. The fourth stage is the classification stage for determining patient's heartbeat cases using the WPCA, which includes three processes, one is the Procedure-FFV for determining the values of the weights according to the qualitative features of the sample heartbeats, two is the Procedure-DPC for providing eigenvectors of U principle components for heartbeat cases determination, and three is the Procedure-HCD for determining patient's heartbeat cases. When the Procedure-HCD has been completed, the patient's heartbeat is then determined. In the experiment, the sensitivities obtained are 95.29%, 93.35%, 92.29%, 79.98%, 91.55% and 90.07% for heartbeat cases NORM, LBBB, RBBB, VPC, APC and PB, respectively. The total classification accuracy is approximately 93.19%.