Dealing with multiple measures etc. can be good, but also misleading. It depends on the accuracy of the model being used. Further if interventions or changes are being looked at over time then it can get sucked into problems such as the hill climbing problem (reference to an artificial intelligence issue) where you can find local optima and minima and think it defines the space. I do agree though that we need to map all the interacting factors and then measure them all. I have long thought that we really need to see if the percentages of patients who have particular issues or fall into particular subgroups can actually do that for many subgroups simultaneously. While the deficits of the current CDC investigation are obvious, it at least allows for many variables to be studied simultaneously. I do not know enough about specific measures though to tell if its adequate. Another problem that can arise is due to the nature of that being measured. Most measures from blood are whole body averages. There are also temporal issues. This tells us far less than I like about what is happening in specific locations, or at specific times. Yet investigation has to start somewhere. Yet static timepoint measures tell us not that much about dynamic issues. One thing that might would be to take a small number of patients, say 10, then track every kind of testing there is over time, say 100 days. That is probably too long for patients to cope, so maybe 10 days would suffice, or every tenth day over a long time frame, or something similar. In other words we need detailed longitudinal measures. Sadly our scientific process is still not advanced enough to deal with real complexity in large complex dynamic processes. Yet we have to start somewhere.