A Time-Varying Seasonal Autoregressive Model-Based Prediction of Respiratory Motion for Tumor following Radiotherapy

Table 1

Tested prediction methods and settings.

Method	Setting
ZOH	None	—
	Samples back length	300 samples
SSA [19]	Dimension number of covariance matrix	250 samples
	Number of largest eigenvalues	18
	Sampling interval	15 samples
KDE [18]	Dimension number of covariate	3
	Length of moving window	300 samples
	Representing value of distribution	Mean
Adaptive SAR [17]	Order of seasonal AR	𝑃=2
	Coefficients of SAR	Φ𝜌=1/2, 𝜌=1,2,…,𝑃
	Order of TVSAR	𝑃=2
TVSAR	Coefficients of TVSAR	Φ𝜌=1/2, 𝜌=1,2,…,𝑃
(proposed method)	Reference interval estimation	(a) Correlation analysis-based approach
		(b) Adjustment with adjustment range 𝐿=5