Celiac here. I went through a not nearly as bad although still 18 month long painful regression, diagnoses and recovery period before getting to my current better than ever state (thanks to diagnoses and fixing my diet). My brother and I wrote a learning algorithm in Go that allows you to log your food and how it made you feel, and then it evaluates that against all of the other things you have eaten and changes the score for that food. After a month or two of data you start to see compelling patterns, and it allows you to isolate single ingredients through continued trial and error. If something results in 10 bad experiences and no good ones, it is probably a trigger food and should be eliminated. On the opposite end, you can conclusively rule out foods by trialing them with a bunch of other ingredients and seeing if they cause problems.

I built it because three people I know went through processes similar to the one you describe and they found the trying foods process very hard to track in a way that made is easy to determine which foods were a problem. I never took it further but the Go app is deployable and I think there is a great service to be built I have just been busy with my day job. I think that if this job goes well I will have enough money to work on this mission full time and at that point I will revisit this project and try to make it more functional and integrate with other medical and health devices and APIs for easier logging and better correlations (ie. eat a food, that increases heart rate 10-20bpm indicating autoimmune response of some kind, that is automatically tracked and the food is scored lower).

I signed up on the website, first name starts with R and time around 21:20, feel free to email me and we can chat.