Letting the EHR help with diagnoses
Q: Since the entire healthcare record is available in electronic form, shouldn't it be helpful if the system would comb through the data and flag diagnoses the doctors have missed?
A: Yes, that would be helpful, but sadly, it's not possible. In order to enable this scenario, there must be at least some data in the EHR system that point to a diagnosis unknown by the doctor and the same doctor must have entered that data. A few exceptions do exist, such as lab data, but the diagnoses that can be made that way alone is too small to matter.
So this idea hinges on the requirement that the doctor is able to enter data into the EHR while at the same time not realizing the significance of that same data set. If he did realize the significance, the diagnosis would already be made and the EHR system itself would have nothing to contribute.
But in reality, doctors don't work that way. We listen to the patient, form a hypothesis, examine the patient to confirm or refute that hypothesis, build a new hypothesis and repeat, until we either get somewhere or fail and refer to someone else. (You can refer for other reasons as well, of course.) That is, the only examinations we do, clinical or by technical means, are directed by our hypotheses. There simply are no examinations made except those we understand the significance of beforehand. We don't randomly poke at the patient from head to toe just to see what happens.
Sometimes we do find something we don't understand within the reference of our current hypotheses, but it's exceedingly rare for a doctor to write down these stray inexplicables. The reason for this is that we first examine the patient, then draw our conclusions, then finally write it all down. Practically everything we then do write down actually supports the conclusions we already have drawn. You hardly ever find anything in a medical record that says "...and the patient has exceedingly thin fingers, but I have no idea what that means." No, it shouldn't be like this, but that's human nature for you. There is very little guarantee of objectivity or completeness in the written record.
The only way a computer could ever make a diagnosis from the medical record that wasn't already made by a real human doctor, is if we had a doctor's assistant that did every conceivable clincial examination of every patient, without understanding the significance or meaning of any of it, and entered into the medical records so the machine could draw conclusions. This exam would take hours for every patient, of course. Or if we had the "Medical Tri-Corder" from Star Trek, it could be meaningful. But we don't have that yet.
There are no comments on this page. [Add comment]