Honestly reading walls of text has gotten me pretty far by just pairing it with lookups. I’d say I misread pretty rarely as far as I’m aware. Not to mention even natives misread so theres a level of shouganainess to it.
After getting to a decently high level where I was coming across fewer than 50 new words per books, I got several audio books and would listen to those while reading or after reading. And then, I would use that to spot when I misread. Honestly though, I think I did this for like 6 books or so and I really only found a couple things I misread like 明日 as あした rather than あす in a certain situation and a couple others I don’t even remember. I think if you’re just diligent about looking stuff up and which reading goes with which situation, you’ll be fine. Like you might be inclined to see 同じ轍を踏む as おなじわだちをふむ (as I was when I first came across it) and feel like you know what it means, but realistically it’ll usually have furigana at some point and even if it doesn’t, its pretty obvious that it has a high chance to be a set saying that you can look up just to be sure of the meaning. In doing so, you should come across the right reading.
Sadly, there is no golden ticket that will help you accomplish this besides reading and reading some more. Not to say that there aren’t some steps you can make to make the transition easier, finding comprehensible input, i+1 and the like, though I personally rather read something I enjoy and is somewhat correct for my level than reading whatever. What I find works the best is just starting at something that you can tackle in a realistic timeframe. If that means starting out with NHK easy news and satori articles, or tweets and using yomichan? Then go for it. Just keep reading and reading. Your brain will eventually sort stuff out. You will have to look up a lot though, it takes at least 12 encounters for a word to settle in your brain, and the more times you see it the more it will settle itself. SRS systems and vocab lists are great to help here in a time spent / results gained level, but to truly cement it there is nothing that beats reading (and reading A LOT).
There are studies out for English languages and the like that claim you need to encounter 11 million tokenizations to learn the most common 9000 words for example, and while this study is done for English languages and we should be careful to draw exact conclusions, you would need to read around 100 books just to cement the 9,000 most common words. But then you’ll have a wonderful base of 9k words that you know VERY well. and thanks to stuff like zipf’s law and stuff, you’ll know most of any corpus there by that point, able to infer most words from context, and only having to look up words occasionally.
Still, 100 books is A LOT! So, get to reading I’d say