Title: Multivariate autoregressive model for ECG signal forecasting

Authors: Sarita Kansal; Prashant P. Bansod; Abhay Kumar

Addresses: Department of Electronics and Communication, Medi-Caps Institute of Technology and Management, School of Electronics, Indore, M.P., India ' Department of Biomedical Engineering, Shri Govindram Seksaria Institute of Technology and Science, Indore, M.P., India ' School of Electronics, Devi Ahilya University, Indore, M.P., India

Abstract: In this paper, multivariate autoregressive modelling is used to analyse the correlation between diagnostic components of an ECG signal. The value of diagnostic components is identified in every beat, and is measured by wavelet transform. The diagnostic components are considered as ECG variables for modelling and it represents the time series signals. The forecasting of ECG variable 'IHR' is evaluated by using multivariate autoregressive model. The model is characterised by different number of ECG variables and past values of each variable. It affects the forecasting accuracy, which is measured by mean absolute error (MAE). The results show that as the number of diagnostic components is increasing in terms of ECG variables, the forecasting accuracy is enhanced by reduction in the value of MAE. The forecasting accuracy is calculated for the forecasting horizon of 80 ECG beats.

Keywords: multivariate autoregressive model; mean absolute error; MAE; ECG variables; AR model.

DOI: 10.1504/IJMDA.2017.087618

International Journal of Multivariate Data Analysis, 2017 Vol.1 No.2, pp.124 - 139

Received: 04 Nov 2016
Accepted: 02 Feb 2017
Published online: 28 Oct 2017 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article