How does Poisson kriging compare to the popular BYM model for mapping disease risks?

Table 2 Performance comparison of Bayesian and geostatistical estimators: mean error of prediction.

Estimators	Lung cancer		Cervix cancer
Arithmetical average	Average	% best result	Average	% best result
Global Empirical Bayes	-0.031	22	0.053	18
Local Empirical Bayes	0.012	14	0.006	28
BYM Model	0.010	12	0.006	16
Point Poisson kriging (adjacent counties)	0.009	14	0.019	22
ATA Poisson kriging (adjacent counties)	0.014	20	0.023	8
ATA Poisson kriging (32 neighbors)	-0.001	18	0.040	8
Population-weighted average
Global Empirical Bayes	-0.026	26	0.012	32
Local Empirical Bayes	-0.011	22	0.001	32
BYM Model	-0.009	18	0.001	12
Point Poisson kriging (adjacent counties)	-0.011	16	0.001	10
ATA Poisson kriging (adjacent counties)	-0.011	4	0.001	8
ATA Poisson kriging (32 neighbors)	-0.016	14	0.002	6

Results obtained on average (arithmetical and population-weighted) over 50 realizations generated for Regions 1 and 2. Poisson kriging was conducted using either adjacent counties (same neighbors as BYM model) or the 32 closest counties in terms of distance between population-weighted centroids. ATA kriging accounts for the shape and size of the counties in the analysis. Straightforward empirical Bayesian smoothers were also applied. Bold numbers refer to best performances. The second column gives the percentage of realizations where the particular method yields the smallest prediction error.

ISSN: 1476-072X