The end-to-end error count sometimes (in rare cases) may indicate problems with the hard disk itself, but generally it is more common when there are problems with connections and/or cables. If these do not provide correct connection and communication between the hard disk and the system, yes, you may experience exactly what you wrote:
- any software running and using the hard disk (or the whole system) can be completely frozen
- after restart the hard disk may be not detected, may be slower - and in most cases "disk error" reported
- sometimes (may be after change of temperature) things may working again.
Changing/replacing the cables is a good idea and in many cases it fixes the problems.
However, it does NOT change the status of the hard disk which detected the issue previously: the change of the cable does not reset / remove the detected problem (so it is still reported - as the hard disk counts all problems in its lifetime).
The most important is that (ideally) after the cable change, there should be no NEW problem(s) detected. To ensure this, it is recommended to perform some testing of the hard disk ( http://www.hdsentinel.com/faq.php#tests ). If these tests show no errors and the status is stable, then in Hard Disk Sentinel you can acknowledge the problem: clear the error counter so then the issue recorded in the past (before the cable change) will be no longer counted and Hard Disk Sentinel will report only possible new problems.
This is described at http://www.hdsentinel.com/hard_disk_case_communication_error.php
(this page mentions a different attribute which usually changes in such situation, but generally the situation is same, except that attribute 184 affected in this case).
Generally you should perform testing first as described at http://www.hdsentinel.com/faq.php#tests to reveal any possible further issues - or confirm that the status is stable. This should help to decide if the hard disk "needs to be replaced" or not.