Unfortunately, it's pretty tough to do it as just one person. All of my models assume you have a population of patients that are tracking their diets. With one person, you probably are best off just tracking your foods, medications, and how you feel (numerically, say [0,10] or something). Then after a while you can do a regression based on how much of a given item you think you have in your system.