Imagine that we are interested to see how the spectrum changes over time during some experiment.

For that, we first need to select `time` as covariate. We then try to describe by means of kinetic models how the intensity of each spike evolves over time.

For each species/\(m/z\), SPIX

- fits a kinetic model to the data,
- compute \(R^2\), i.e.Â the fraction of variance of the data explained by the model,
- compute the \(p\)-value of the test used for testing if the model reduces to a constant model (no trend). Then, a small \(p\)-value indicates that we are confident that the intensity of the spike, i.e.Â the amount of the associated species, changes over time.

For each species, the ``bestâ€™â€™ kinetic model is selected among a library of 7 models with different shapes:

- models \(A_1\) and \(A_2\) decrease over time
- models \(B_1\) and \(B_2\) increase over time
- models \(C_1\) and \(C_2\) increase and then decrease
- models \(D_1\) decreases and then increases