Is the only reasonable solution to schedule a "human cron" once a week
to look
at needed updates? Ouch.
A middle-of-the-road approach is to have a machine or VM where you can test things, perhaps the one you use as your own desktop or for development, where you have all the packages installed that the other systems use. You can 'yum update' this one frequently, noting what packages are affected and that everything still works after a reboot (for things where that might make a difference).
I use a VM set up this way with the following crontab:
# check for yum updates every 12 hours 5 0,12 * * * root /usr/bin/yum -q check-update 2>/dev/null
so I get an email whenever there's any updates due. I can then evaluate, test, and (perhaps) schedule a time to manually update the production servers.
The yum-updatesd package does all of this. Its config file is pretty simple and has your choice of whether to download, whether to install, and where updates should go.