Predicting length of stay from an electronic patient record system: a primary total knee replacement example

Table 3 Akaike information criterion for the models

Model	AIC	Diff AIC	Model likelihood	AIC weight
Negative binomial	9170	0	1.0	1.0
Poisson	9668	498	0.0	0.0
	Total model Likelihood		1.0

ISSN: 1472-6947