Generally troubleshooting to know the cause of the failure needs to be done...
It can be dangerous to automate "fixing" the service when you do not know why it isn't responding the way you expect it to. For example, if you force an automatic restart on a 15 minute processing delay, you have compounded the processing delay with the time to restart and reload the caches. If the cause of the processing delay is data overload from the AMDs, the forced restart will only make the delay worse.
If the cause is a corrupted SQL database, forcing the CAS to run anyway will further corrupt the database.