It's tedious, but you can avoid rebuilding it from scratch.
You can troubleshoot by feature type:
A) Turn off all the quiz questions, and quiz results; republish, and test.
If it works, you've identified the part of the course needing your attention.
If it doesn't work, you need to turn those features back on, and turn the next block of features off.
Turn off all the learning interaction slides; republish and test.
If it works, you've identified the part of the course needing your attention.
If it doesn't work, you need to turn those features back on, and turn the next block of features off.
Repeat again for Video, animations, etc.
Or, you can troubleshoot by brute force, treating all the slides as equal:
B) Break the whole course into blocks of slides, maybe divide by 5 or 10. A 100 slide course could be tackled in 10 blocks of 10 slides, or 5 blocks of 20 slides. You'll possibly need to publish each block, so deciding for more blocks means more round trips.
Turn off the first block, republish, and test.
If it works, you've identified the part of the course needing your attention.
If it doesn't work, you need to turn that block back on, and turn the next block of slides off.
Turn off the second block, republish, and test.
If it works, you've identified the part of the course needing your attention.
If it doesn't work, you need to turn that block back on, and turn the next block of slides off.
Turn off the first block, republish, and test.... lather, rinse, repeat!
Once you've identified the troublesome block, you need to test for which slide might be the problem.
This type of question is why I tend to build features onto a course rather than working from front to back and completing each slide. As I build features I create versions, publish and test.
I cycle through test environments by testing: locally, to a local DEV webserver, to a PILOT LMS; finally to a PRODUCTION LMS. At each stage I add features and test what was added.