I don't think OBS is going to work for that. Not by itself anyway. If it could choose channels from a device, it probably would, but it doesn't.
Instead, it takes the entire device and mixes it down to whatever channel count OBS is set for, and THEN gives you that mixdown to play with. If you're set to stereo, for example, then it'll take your 4-channel device as two stereo pairs. Channels 1 and 3 mix to the left side, and channels 2 and 4 mix to the right side. THEN you can work with that stereo mixdown...which is not what you want.
I found that out with a digital mixing console, that had an 18-channel USB interface. I was set to stereo, and it took those 18 channels as 9 stereo pairs. Odd channels went left, even went right, and no way to do anything different. Which was a problem because the first 16 channels were the console's raw inputs that I didn't want at all, and I only really wanted channels 17-18 as my finished mix to pass through. In that particular situation, I wasn't using the USB connection for anything else, so I could go into the settings on the console side and make everything silent that I didn't want, but you can't always do that.
If you were using a 2-channel interface, I could tell you to have two copies of it in OBS, mono both, and pan them hard to either side. That works because the pan/balance goes before the mono switch, which is backwards from pretty much any professional thing, but it makes this particular trick work. But I don't think it works so well for anything beyond 2 channels of raw input. OBS's panner seems to be hard-coded to stereo, regardless of the actual channel count.
You ***MIGHT*** be able to get what you need, if you set OBS to 4 channels of audio (4.0, or "quadraphonic"), and don't try to remap or route anything. But if you need to engineer all 4 channels into a stereo or mono main mix, that won't work either.
I thought briefly about Voicemeeter, but it actually has a similar problem as OBS does. It doesn't downmix (which is already an improvement over OBS), but it doesn't separate either into independent mono's.
So having exhausted everything else I can think of, I think you need a for-real DAW. Bring your live inputs into the DAW, and do all of your audio stuff in there. Not OBS.
If a DAW is too much, then you might see if you can record a single 4-channel audio file and then split it with Audacity. Export your 4 mono's from Audacity, and go from there. That would not work live, but since you say you're recording, it might.
You might even do the 4-channel recording directly in Audacity, split and export, and not use OBS at all for this.
