I'll speak from my experience about how medical testing occurs here.
GP's are unattainable for most. Specialists have months or years long waiting lists.
Any doctor I do get to see looks for the shortest exit, so if a syndrome says muscle pain then they prescribe for muscle pain. If you're lucky, the doc has the time and there is suspicion there might be a problem they run a screening test. Note the chances of seeing the same doc is near impossible and the medical records are somewhat localized to the doc/clinic. This doesn't work for diseases/consition that require a GP quarterback and seem to be resolved in one visit.
The screening tests are often just that. To save money, any secondary tests are not done of the screening fails. If an ANA test is done and comes back negative then further tests are kicked out by the lab. Report comes back to the GP, nothing is flagged so no follow-up is done. This also happens here with thyroid tests too. Even is there is a box for the GP to override, it is unlikely done. Often the specialist has to specifically ask. GP has to have strong evidence to move the patient forward. The result is that rare diagnoses become rarer.
Summary research papers look at large amount of data and conclude that there is no correlation, so rare diseases get rarer.
An AI is "trained" in so much data that they produce "conventional wisdom". Getting it to produce a plausible radical alternative is difficult. For instance, homosexuality stopped being a disease when it was removed from the DSM. We no longer talk to parents about the "medical risks of homosexuality " but there are many current stories about "autism risks"
GP's are unattainable for most. Specialists have months or years long waiting lists.
Any doctor I do get to see looks for the shortest exit, so if a syndrome says muscle pain then they prescribe for muscle pain. If you're lucky, the doc has the time and there is suspicion there might be a problem they run a screening test. Note the chances of seeing the same doc is near impossible and the medical records are somewhat localized to the doc/clinic. This doesn't work for diseases/consition that require a GP quarterback and seem to be resolved in one visit.
The screening tests are often just that. To save money, any secondary tests are not done of the screening fails. If an ANA test is done and comes back negative then further tests are kicked out by the lab. Report comes back to the GP, nothing is flagged so no follow-up is done. This also happens here with thyroid tests too. Even is there is a box for the GP to override, it is unlikely done. Often the specialist has to specifically ask. GP has to have strong evidence to move the patient forward. The result is that rare diagnoses become rarer.
Summary research papers look at large amount of data and conclude that there is no correlation, so rare diseases get rarer.
An AI is "trained" in so much data that they produce "conventional wisdom". Getting it to produce a plausible radical alternative is difficult. For instance, homosexuality stopped being a disease when it was removed from the DSM. We no longer talk to parents about the "medical risks of homosexuality " but there are many current stories about "autism risks"