I don't know how you intend to do your data analysis, but... I have had an idea for a smartphone app on my Big List for years. This app would implement lag-seq

I don't know how you intend to do your data analysis, but... I have had an idea for a smartphone app on my Big List for years. This app would implement lag-sequential analysis (see for example http://www2.gsu.edu/~psyrab/references.pdf http://www2.gsu.edu/~psyrab/references.pdf). I haven't actually used it (which is part of why I have procrastinated this one), but my understanding is that LSA is basically a chi-squared analysis with a time function. Chi-squared compares expected vs. actual incidence of something. So LSA would look at expected vs. actual incidence of X happening following Y. The prototypical test would be to prove/disprove that chocolate milk makes kindergarten kids more violent. But you could (as I envisioned it) generalize it to sniff out all kinds of factors that might systematically precede migraines or whatever. You'll also need to explain to people that a negative result might not be a reliable one unless you get a lot of data and record all variables reliably. I'm not sure how much my app idea overlaps with yours, but I will cheer you on in any case.