These are not necessarily issues. What matters more is the true execution time of the relevance of these actions / relevance statements, but I believe the health check report only looks at the character length of the relevance statements, not the true execution time.
If any custom items get flagged by this, it may be a good idea to take a look at them to see if they can be optimized.
There is something you can enable on endpoints to measure the relevance evaluation time of actual fixlets/tasks to identify any expensive relevance, but it is not the easiest thing to investigate.
Are you having issues with client report performance? What does the evaluation loop inspector tell you about the eval loop on your clients?