I am not using Dgraph for healthcare at the moment. The hospital where I work is all about on-premise tech and I have only been there a little over 3 months, so not ready yet to push for Dgraph, plus it does not yet claim the security that is needed for ePHI. It is working towards this though. IMHO compliance is more about how it is implemented than the actual service itself, but there are a lot of items to check of to ensure compliance however you go about it.
I am using Dgraph right now for MissionBase (not publicly available yet). I gave a talk about it on the last Dgraph Day
(Not my finest picture, lol)
My data is only at 0.21Gb per my Cloud dashboard which is kind of crazy for how much data is actually there. I have ~1,140 predicates spread across 140 types.
At first (June 2020) I tried to deploy Dgraph myself in AWS, and that was difficult as I am not a AWS engineer by any means. I did get it up and running, but not comfortable enough with server management to feel safe to do it all myself. Dgraph launched Dgraph Cloud (then called Slash GraphQL) and I was one of the first users, and have been using it ever since. I am currently just using the Dedicated plan and do not have any current major issues. I say no major issues, because there is always a learning curve and the little things here and there to overcome, but very feasible. I am not the smartest guy in the forum, just a redneck from Southeast U.S. and self-taught. I believe most anybody can easily learn and use Dgraph if they are dedicated enough. It is easier to learn than many other programming/db query languages IMHO.
So, I am not the biggest user on the network obviously with less than 1Gb of data. Here is something that may interest you:
I know there are bigger graphs in production other than mine and maybe others will respond here too 
And I don’t know if you do Discord or not, but here is an invite: