Subject: My thoughts on Issues 5 and 9 that were assigned to me...
Frankly, I’m not entirely sure my thoughts here are going to stay precisely on these two issues, but let me throw them out anyway…
To set the set the stage …If we look at a simple medical symptom scenario…
A sick person calls a doctor and says, “I have a really bad headache.”
The doctor responds, “Take two aspirin and call me in the morning.”
Analyzing this from a SAF perspective, the portion of this scenarios we don’t have covered is the “call me in the morning” portion. We have discussed, of course, that the patient can always report a new symptom (“The headache went away.” “The headache stayed the same.” “The problem got better but I still need help.” “The headache got worse.” “I died.” Etc.), but since we don’t have ANY notion of a “case” in the standard (or a linking to a prior symptom), it’s not clear to me that a diagnostician can efficiently determine that a prescription actually worked (which also means that it is difficult for it to decide which symptoms are not longer applicable).
A related issue is illustrated by the contrived medical example…
A person calls a doctor and says, “I got hit by an ax this morning AND I have a really bad headache.”
The doctor responds, “Remove the ax AND then take two aspirin and call me in the morning.”
At first glance, it appears that we have the first part of this covered, since a symptom source can certainly report two different symptoms, and, in theory, these can be correlated with a syndrome. However, we don’t have any (formalized) way in the framework to let the reporter STATE that it believes two symptoms are related, which is probably very valuable information to a diagnosis.
Likewise, we don’t have a way to associated two prescriptions. Yes, we can create a compound prescription (“Remove ax and then take two aspirin”) but this seems to me to create unnecessary complexity.
Not to expand the scope any further, but we also don’t have any clean way to have the doctor “gather more information” about a condition:
News Feed: Crazy Ax-man escapes from jail!
Patient: I have a headache
Doctor: Did you, by chance, get hit by an ax this morning?
Patient: Actually, I did.
As much as I like the simplicity of our model (simple reporting of single atomic symptoms, and simple issuing of atomic prescriptions), I wonder whether we don’t need to have some kind of an explicit feedback and linking mechanisms in the model.
Patient: I am calling about the headache I reported yesterday. I took the aspirin. My headache is now worse and blood is now gushing out around the ax.
Doctor: Are you still having the headache you reported yesterday.
Patient: Nope. I took the two aspiring and my headache went away.
Now we might be getting in to “case management” at this point (Frankly, I’m not sure I understand the domain of case management), but maybe not. At a minimum, I think we may want to consider how we would interface with a case management standard for a scenario like this.
However, we might be able to support all of these scenarios in the absence of case management with something like the following:
1. Some notion of a link, within a symptom, to a related symptom.
2. Perhaps this is the same as item 1, but some notion of that this is a follow-up report to a previous symptom.
3. Some “inquiry” interface back in to the symptom source. Perhaps with 1 and 2 above this could be handled via prescriptions.