Pharmacokinetics is the study of the time-course of drug and metabolite concentrations in the body.  Decades ago, pharmacokinetic data analysis was limited to join-the-dots style non-compartmental analysis (NCA), which provided descriptive parameters like Area-Under-the-Curve (AUC) or Cmax.  Then, insight into the drugs behaviour would be inferred from these simply derived numbers and a knowledge of the relevant physiology.  NCA still plays a role in bioequivalence studies and quick-and-dirty analysis, however the utility and impact of pharmacokinetic data has increased massively since the arrival of the population approach.

The population approach was developed by Lewis Sheiner and Stuart Beal in the late 1970's.  It's central theme is to separate the genuine differences between individuals from the random noise in pharmacokinetic measurements.  As pharmacokinetic models are nonlinear, this requires computer estimation of a nonlinear hierarchical model.  Fitting nonlinear hierarchical models was beyond the software of the time (and remains challenging even today), so Sheiner and Beal developed the NONMEM software.  This innovative package made more advanced PKPD analysis possible.  The population approach produces more accurate estimates of physiological parameters, in particular variability among individuals, and the final model can be used to simulate future trials or scenarios. The population approach was a breakthrough that was both profound and massively influential, and perhaps even worthy of a Noble prize.

For simple datasets, the benefit of population analysis may not be immediately obvious, compared to the relative ease of automated non-compartmental analysis.  However, as soon as datasets become more complicated, for example with complex pharmacokinetics, multiple studies or the addition of just a single biomarker, the population approach is essential to fully analyze the dataset.