modelmodel is a python package for creating programs to do detailed, and if we're optimistic, quantitative analyses of BOLD timecourses - real and simulated.

The focus is on model-based (i.e. parametric) designs.

It is still very much a work in progress.

The great work of statsmodel project and especially patsy made this package possible and powerful.

I've written the analysis tool I have always wanted.

Specifically:

• To seamlessly intermix real and simulated BOLD data, allowing for detailed analysis and assumption testing.
• Designed to make specifying a model-based design trivial; It's the only simulation environment focused on model-based designs
• If you need to integrate computational model parameter fits, it can do that too.
• Has builtin access to over 500 anatomical ROIs from 8 separate atlases.
• It's trivial to add your own (functional) ROIs.
• Model-comparison is the default approach, with AIC the favored statistic. But BIC, F-values, and other others are supported.
• It's very easy to swap in sophisticated regression techniques in place of OLS. Any regression method from statsmodels will do.
• It's the only fMRI simulation environment for the python programming language, as least as far as I am aware (if this is wrong, please let me know). Note: There are very nice systems for R (neuRosim) and MATLAB (simTB - which looks quite fantastic these days).
• Designed as a programmer's analysis package first. I wanted a clean but powerful and pythonic way to do a fMRI analysis. So that is what I tried to write.

That said, I really hope it is useful to you.

This is not a beginners tool.

• I expect you have a solid grasp of fMRI analysis methods.
• I expect you can program in python.
• I expect you can preprocess the data elsewhere. Data must be in Nifti1 (and MNI152/352 space, to use the ROI features).

If any of these expectations are not met, you are going to have a bad time.