# Non-Linear Regression Analysis of Pharmacokinetic Data Individual Data and Population Analysis

previous | next

### Student Objectives for this Chapter

• Understand the reasons why models are developed and used
• Understand how models can summarize or 'compress' data
• Understand how models can be used to study pharmacokinetic mechanisms
• Understand how models can be used to predict concentrations or dosage regimens

• Understand the use of formulas as 'Mathematical models'
• Understand the criteria of least squares
• Understand how parameter adjustment changes the fit to data

• Understand the use of computer programs such as Boomer for non-linear regression analysis of pharmacokinetic data
• Consider Bayesian analysis of clinical data
• Understand the use of computer programs such as NONMEM for non-linear regression analysis of population pharmacokinetic data Figure 25.1.1 Diagram Illustrating Data Analysis Paradigms

With no patient data but patient information no data analysis is required. Dosing calculations are based on nomograms, package inserts or other published information. If a few data points are available from a single patient and population parameter values for the drug of interest it may be possible to perform a Bayesian analysis. This analysis method can combine these two pieces of data. If only a few data points are available for each subject but data are available for many subjects a population analysis may be possible. This can also provide population parameter values with measures of the uncertainty in these values. More data in one subject allows 'traditional' non-linear regression analysis using graphical methods or a computer program. With more data from multiple subjects a population analysis is again very useful. Alternately a two step approach of analyzing each subject's data separately using non-linear regression analysis and combining these results may be applied. A third approach might be to combine the data from multiple subjects in a naive pool analysis.