You will need to do much more assessment of your endpoints failing since this is an issue with your endpoint 802.1X supplicant behavior and not ISE.
Are these large numbers of random 5440 failures truly random and from unique endpoints? Or... are they repeated at random intervals from a subset of endpoints in a specific location or through one or more network devices or even talking to a specific ISE PSN node? You need to narrow that down and troubleshoot from there. If it is truly random, revisit your supplicant configurations' timeouts which I assume is globally applied to all of these endpoints and ensure it is following our best practice timeout recommendations (7 second timeout with 3 retries).
Most likely culprit is a misconfiguration (ridiculously low timeout) or network latency (no response from ISE to endpoint before timeout) resulting in the endpoint trying again. Ultimately, ISE does not know why the endpoint did what it did, it just knows the endpoint initiated a new RADIUS session while ISE was still working on the old one.
From List of Cisco ISE Syslogs :
Message Code: 5440
Severity: WARN
Message Text: Endpoint abandoned EAP session and started new
Message Description: Endpoint started new authentication while previous is still in progress. Most probable that supplicant on that endpoint stopped conducting the previous authentication and started the new one. Closing the previous authentication.
