This is one of the things I find most compelling about a tool like OpenEMR, although I’ve never had the need to use it this way. While I think this has been done a few times before in various ways and for various reasons (see e.g. OpenEMR on Raspberry Pi 400). I think what I’d be thinking about is this:
If this is a disaster scenario, I’d run OpenEMR independently on each laptop, and worry about syncing/merging data later, when you can be sure of stable network and power. It’ll create some issues, such as two different patients having the same id on different laptops, but those can be overcome easily enough later, when there’s time. (Or, if you have time to prepare, offset the patient ids in each install by a large enough number.) I’d also give some thoughts to ways to manage backups, either to a local sturdy drive and/or to the cloud if internet exists, but is intermittent.
If you arrive and find out that power and network are not a problem, then you can always expose one of the laptops on a local network and make all the roamers use it instead of their local installs.
At a high level it seems to me that it’d work great.