Fig. 8
From: Nonlinear mixed-effects modelling for single cell estimation: when, why, and how to use it
Analysis of STS, NONMEM, and JLH dependency of number of data sets using Model 1. The figure shows STS (red line), NONMEM (black line) and the JLH approach (blue line) precision in the combined parameter and noise estimation problem, with respect to number of datasets. In the JLH approach, a joint likelihood without postulated parameter distributions has been used. a the results from the parameter estimation as the normalised sum of the absolute values of the deviation from the true parameter. b the results from the noise estimation of true noise level (green line). Estimates of both parameters and the noise from both NONMEM and JLH are closer to the true values than estimates from STS. Also, it is clear that NONMEM converge faster towards the true parameters and noise with respect to the number of data sets, when compared to JLH