Table II.

Bias (MRER) and Precision (MARER) of Estimated Fixed and Random Effects for the Second Model/Sampling Design Scenario: Two-Compartment Linear PK Model with Sparse Sampling Design

Methods	CL	V c	Q	Vp	ω 2 CL	ω 2 Vc	ω 2 Q	ω 2 Vp	σ 2
MRERa
N MC = 1000	−0.89	−1.1	−1.0	−1.1	−4.8	11	50	−12	−0.81
N MC = 10000	−0.65	−0.15	−1.8	−1.5	−4.0	22	52	−6.7	−2.5
MARERa
N MC = 1000	2.0	3.7	6.0	4.7	12	22	53	21	5.7
N MC = 10000	1.9	4.5	4.2	4.2	12	30	54	23	5.4

MRER mean percent relative estimation error; MARER mean percent absolute relative estimation error. Total number of simulation trial used for MRER/MARER calculation = 100; number of subject for each simulation trial = 200; N _MC the number of Monte Carlo parameters set used to compute the E-step of the MCPEM algorithm; CL population (typical) clearance; V _c population (typical) distribution volume at central compartment; ω ² _CL population variance of CL; ω ² _Vc population variance of V _c; σ ² variance of intra-individual proportional error model

^aBoth MCPEM_GPU and MCPEM_CPU produced identical results