2nd Level of DCM after RFX BMS

We want to use dynamic casual modeling to evaluate effective connectivity on glioma patients, but I’m really new to this. I read a[guide paper](https://doi.org/10.1016/j.neuroimage.2009.11.015) and it suggest that it is more suiteable to use RFX BMS for model selection. But I’m confused about the relationship between RFX BMS and PEB in the 2nd level DCM of SPM, is it suitable to use PEB framework to do the 2nd level analysis after RFX BMS or what is the proper way to do the 2nd Level of DCM after RFX BMS?

Best wishes,