High-dimensional ARMA model identification and its application to healthcare picture smoothing using a forgetting factor

In this paper a set of formulations of an N-dimensional (ND) autoregressive-moving average (ARMA) model identification method, and a two-dimensional (2D) forgetting factor approach in time-series modelling, is developed. An optimum estimation and prediction approach in healthcare picture smoothing based on a 2D ARMA modelling, has been implemented; and satisfactory results have been obtained. Our approach indicates the desirability of accurate statistical modelling of high-dimensional or periodic digital data.