[tsoukase]

Very rare disease, expected to be misdiagnosed as psychiatric. I admit I would (neurologist here) But you were luckily treated by an excellent neurologic center.

The lesson learned is that there are rare diseases (<1/10.000-100.000) but as they are so many, they form an important minority next to common ones (1/100-1000). Just don't forget them when data don't fit well. Such estimations is a hard dexterity of doctors that cannot be rivalled by AI.

[burntsushi]

I asked my wife about it since she has more context on the chance encounter with a neurosurgeon who was ultimately responsible for the transfer into Brigham and Women's. She said there were two things of note that led the neurosurgeon to step in (according to the neurosurgeon):

Decompensating at an increasing and alarming rate, not typical from a psych perspective. Also not in a healthy young male with no history. My wife was able to substantiate this claim with my detailed timeline. (I had some emails and wrote a lot of stuff down, as is my nature, during this time.)

My left eye was "squinty."

[burntsushi]

> My wife was able to substantiate this claim with my detailed timeline.

Sorry, my bad here. The timeline wasn't mine. It was Kait who had built it and took on the burden of sharing it with everyone she possibly could.

[rendaw]

I don't mean direct this specifically at you, but aren't there databases that can look up diseases by symptoms? Are doctors not trained to consult those, maybe after ruling out common causes? Why is forgetting relevant in this context?

I'm asking because I've had frequent encounters with doctors whose process seems to be, literally, "remember if there's anything like the described symptoms that I learned in medical school" which, if they were somewhat older, was probably 30+ years ago.

[tsoukase]

Esp in mental health there is a large overlap of symptoms between diagnoses. So doctor's experience is needed to get the right diagnosis. No database and, hell, no AI assistance yet.

It all boils down on how deep a doctor goes diagnostically: for common diseases (1/100) an interview is enough, for rarer (1/1000) an MRI/blood test are needed and for very rare you need extensive and expensive genetic and immune exams. Because of cost, the 'art of diagnosis' is to be able to filter the cases that should be referred downward. Classic type I/II error due to binary misclassification happens all the time.

[burntsushi]

After my abnormal brain MRIs, my understanding is that the diagnosis came down to either some kind of encephalitis or multiple sclerosis. But test results have to come back first. I was even set up with a follow-up with an MS specialist. Once my anti-NMDA receptor encephalitis test came back positive though, that follow-up got cancelled and replaced with a neuro-immunologist that specializes in these sorts of things.

And even if I did have an MS, there are various sub-types that require going through something quite complicated call the McDonald criteria: https://en.wikipedia.org/wiki/Diagnosis_of_multiple_sclerosi...

[pygy_]

Psychiatrists are tourist guides for the Paris catacombs who try to get around using an underground map.

When you do cluster analysis of the symptoms of psychiatric patients, the empirical clusters don't match the DSM entries.

Schizophrenia is at least 8 distinct diseases (see GWAS), so orphaned that they don't have a name.

Psychiatry is an epistemic mess, and medicine uses it as a garbage bin for the patients it doesn't understand (because blissful ignorance is the norm over there).