Thanks OP for the post. I'm curious as to how this handles/you plan to think about "conditional" relations between nodes. E.g., for the Ginger <-> Nausea connection, is there a way to describe this link itself as being enriched in a subset of patients, such as just pregnant people? It looks like you derive association scores via text scraping then turning some text analysis into postive vs negative associations -- I'd note that lots of studies include inclusion and exclusion criteria by which this could be explored.