i finished my phd quite a few years ago. My topic started out as signal processing applied to speech, but morphed into signal processing applied to bioinformatics because it was easier to publish.
it was an extension of undergrad, electronics engineering.
a day was either reading papers or writing matlab code, or waiting for matlab code to finish running. Pretty much everything was either reading research or working on future research. If I tried 20 ideas, one might work well enough to get a paper out of. Then a whole lot of testing to flesh it out, then writing the paper and editing and typesetting a creating images.
I was in a lab so there were other people to talk to working on similar things, that helped a lot.
most useful skills i think are just the research process, how to go from undergrad to expert in a topic. how to interact with the academic community.
I enjoyed the process of learning so much. Not so much the process of writing papers.