In the Defensive Programming section, there is a test for patient_normalise that checks a ValueError is raised when negative values are passed to the function:

with pytest.raises(raises):
    npt.assert_almost_equal(patient_normalise(np.array(test)), np.array(expected), decimal=2)

This could be simplified to:

with pytest.raises(raises):
    patient_normalise(np.asarray(test))

Also, it's best to check the specific error message being thrown:

with pytest.raises(raises, match='Inflammation values should not be negative'):
    patient_normalise(np.asarray(test))

And there is a check for whether raises is ValueError or None:

if raises:

However, it's better to always explicitly check that a variable is not None:

if raises is not None:

It makes no difference here, but other times some statements (e.g. []) could evaluate to False and lead to the wrong code block executed.