In our use-case we have Processes which are not running all the time, to it's normal that they become "unavailable" by design.
On the Processes->RandomProcess.exe Page there is this beautiful "Events" Graph which would be nice if we could use them for custom metrics. Or from the Graph in the "Application & Microservices" - "Profiling and optimization" - "Crashes" section.
This can be achieved with Event API, via a developed script to make an Event API call periodically and limited it to only crash events. In my environment I am monitoring a large scale of servers with more than 600 applications, most of the crashed events actually come from unimportant processes which only executed at ad hoc basis. To filter out the noise, I will proceed further to verify the crashed process uptime using the time-series CPU usage metric to filter out the false alarm.
Please refer to the attached picture below to understand how it works, hope it helps.