Kinetics Lecture 9-23 Part 10

Front 1

data analysis approach

fit the data by a proper equation
for a 3 compartment model:

Back 1

c= a1*e^-b1*t + a2*e^-b2*t + a3...

Front 2

data analysis: fitting procedure

pharmacokinetic models are nonlinear with respect to optimized parameters, ie:

Back 2

the dependent variable c cannot be described as the sum of functions of independent variables (time t) multiplied by the optimized parameters

Front 3

initial estimates can be obtained by the:

Back 3

methods of risiduals (peeling, feathering)

Front 4

the best model is selected using statistical criteria that include the quality of the fit:

Back 4

--x^2, sum of squares of deviations (the smaller the better, but cannot be smaller than the experimental error. does not consider the number of parameters)
--the correlation coefficient r (percentage of explained variance = r^2. the closer to 1 the better)
--the F-test criterion

Front 5

another criterion: the deviations between the model and the experimental points need to be distributed:

Back 5

homogeneously

see slide page 1

Front 6

if several models are of equal quality, the simplest model (containing the smallest number of optimized parameters) is selected. The rule is called:

Back 6

Occam's Razor

Front 7

data analysis: summary

Back 7

- method of residuals used to get initial estimates of the model parameters
- the parameter values are refined using nonlinear regression analysis
- for single dose, number of exponentials = number of compartments if the exponents are sufficiently different.