wrt the driving concern: stomach aches, back pain, et al are often psychogenic, and institutionalizing generalized recording to find broad associations are unlikely to be helpful.
I would recommend being more of a scientist: identify a specific issue, canvas the literature for some causal hypotheses based in what we already know, and then design the experiment. Or, ask a professional, i.e., a doctor.
Probably the most fuzzy but prevalent driver is inflammation or an overactive immune system. Blood tests will say if it's hyper-active, and panels can tell you if you're allergic to everything from gluten to types of seafood. (You can order them directly from Quest Diagnostics for ~$200-$600 in the US.)
wrt system, I use markdown both for journaling and for analysis. I reduce everything to headers, bullets, and tables. Then I use full-text search from the keyboard to find things, and parsing to process, e.g., searching for data under a given set of topics.
For document granularity, use some combination of workflow and deliverable. My WIP stream has preliminary evaluation in time order, but I glean results to deliverables -- like lab notes vs. papers.
For details, you end up with your own tag/format system. e.g., most of my urls are something like
- [name - description in {language or system} from {source} on {plaform or date} from {domain}](url)
It reads normally, and allow me to find e.g., matching items from a source.
Obviously, the killer app here is a personalized voice assistant, which chunks your information, instead of your having to fit into some system. I'd pay for that.