I'd be interested in a prospective study that tries to capture data on patients as the severity of their symptoms undergoes long term shifts.
Anecdotally, it seems like some patients experience a level of improvement (though not necessarily recovery) during a window of 3 to 5 years after onset.
It would be interesting to have a lot of biological data on a group of patients at 2 years after onset and then follow up on those who improve prior to (say) the 6th year and see how their post-improvement data compares to their earlier data.
The idea would be to compare data from the same person over time as their severity changed. Obviously, you could also capture changes which might occur as severity increased. That might be harder because a lot of the decline might happen in the first year - before a patient even qualifies for a diagnosis of ME/CFS.
Anyway, it may be naïve, but I'm thinking that if you want to find clues about what's causing the problem, look at what (if anything) is changing when patients get better or worse.
The answer might be something non-intuitive, like if some readings got objectively "worse" by normal standards when some people improved - which might indicate an adaptation to the disease.
It sounds like something similar the OMF Melbourne study is doing - although not to the length and extent you propose.
In its first project, the Melbourne research team will characterize the biology of people with ME/CFS by continuously monitoring their health data and sporadically sampling and analyzing their blood and urine over the course of a year. Data will be analyzed to identify characteristics of disease severity in individuals, then compared across patients to identify patterns — helping reduce the complexity and length of the personalized medical approach.