Machine learning links brain connectivity patterns with psychiatric symptoms | Hacker News Reader

Machine learning links brain connectivity patterns with psychiatric symptoms

healio.com

113 pointslaurex21 comments

The concept sounds perfectly acceptable and promising: documenting brain activity to aid in the diagnosis of mental illness, but this treads on some touchy issues. Specifically, what is and isn't considered "normal". The article notes that...

"Unlike other branches of modern medicine which use biologically-based tests of disease, psychiatry still relies on patient reports and physician observations".

That is true. But unlike other branches of medicine, psychiatry doesn't try to fix people that aren't struggling. We go to the doctor, who then tells us our BP is too high and we need to exercise more. We don't go to our psychiatrist for such checkups. A doctor knows what good and bad blood pressure look like, and will seek to move us from bad to good. Today, psychiatrists rarely ever examine, let alone treat, people until their mental state is negatively impacting their lives. They treat the clinical case. They don't go around telling people they are sick in the way medical doctors look at blood pressure. A brain scan, a "biologically-based test" opens the door to treating people who aren't yet a clinical case. I am sure that such scans would aid in the diagnosis and understanding of mental illness. But by expanding the diagnostic process, these scans would also open up a huge new market for the psychiatric treatment of non-clinical cases.

> But by expanding the diagnostic process, these scans would also open up a huge new market for the psychiatric treatment of non-clinical cases

But is that a bad thing? You are free to ignore their treatment suggestions just like with doctors. My doctor actually did tell me to see a therapist based on some test he had me take.

I suspect it would be better if we could prevent issues from arising in the first place but we don't understand enough to even know whether that's possible. Still, having better biological markers may make that possible no?

I can understand your concern but I like to think we can trust the medical community to make good judgment regarding the tradeoffs between invasive treatment and risk just as they already do.

Uhm. A large number of people, especially children are subjected to psychiatric treatment that they do not request. Courts routinely order treatment for pre- conviction defendants. Schools routinely send under performing or 'difficult'students to psychologists for treatment. These are non- consensual relationships and a clear violation of the code of ethics. These practices taint studies the victims are included in. These forced relationships breed hostility for the field and further isolate the very people supposedly 'getting helped' by these systems.

Courts only order treatment for defendants pre-conviction if they are determined to be mentally incapable of standing trial. This is a very basic standard that simply requires that the defendant understands what a judge is, what a jury is, what the charges are, etc. Are you actually suggesting that the ethical thing to do is to try and potentially convict people too mentally ill to understand what is going on?

Also, I'm not aware of any public school, at least in the US, that sends students to psychologists for treatment. Schools don't even employ clinical psychologists or psychiatrists, and they can't require that a student go to a psychologist to remain at a school.

Disorders which patients don't want to treat are called egosyntonic. Examples: narcissism, anorexia, OCPD.

That the patient, on their own, does not want to treat. But when those disorders negatively impact their lives they become a clinical case that can be treated. Either some legal proceeding causes them to opt into treatment (ie to keep/win custody of a child) or those around the patient insist they seek help.

Hasn't it been reported that psychosis is caused by a lack of connectivity in the brain? I seem to remember someone saying that there had been a breakthrough in discovering that certain brain cells were pulling double duty as immune cells and overzealously pruning connections, causing psychosis.

If over-connectivity is the source of all these problems then isn't that a boon because it must be simpler to cut connections therapudicly than to create them

Those are two different kinds of connectivity.

The first is about synapse density, the second about how strongly different parts of the brain are linked together (relative to one another).

So a brain with a low average synapse density may still have some areas more functionally connected than the average population.

The default mode network and the attention-related fronto-parietal networks are usually disconnected functionally, they inhibit one another.

See https://en.wikipedia.org/wiki/Dynamic_functional_connectivit... for details

There seems to be some thought that it could be under/over connectivity, based upon conditions and symptoms (one of a million theories). I don't know necessarily if more connections can be solved so easily - there's the question of how to keep them from growing back in the same pathological manner, and then also which ones to cut. "Well, this neuron either kills your fear of dogs, or your awareness of the number 5."

also "psychosis" could be better defined. The visuospatial perceptual+ memory errors of an Alzheimer's patient probably very different neurophysiology from hallucinatory phenomena of a schizophrenia patient, for example

If psychosis was due to irreversible brain damage per se, then it wouldn't be possible for it to come and go. Episodes might do cumulative damage, but the difference between being in an episode and not being in an episode can't be caused by something irreversible.

I don't think the root cause(es) of psychosis have really been determined yet. I think it's fair to say that since many drugs induce immediate psychosis, it's unlikely that pruned connections are the only cause.

It’s pretty clearly D2-like receptors being overexcited, at least as far as cause and effect. Hence psychotic symptoms in stimulant users and a remission of psychosis in schizophrenic patients when on D2-blocking drugs.

D2 receptors are undoubtedly tied to many forms of psychosis, but their being the root cause is not yet proven.

Psychosis resulting from meth withdrawal has always suggested to me that it is an under-connected problem because presumably what happens during withdrawal is widespread "disconnection" resulting from a large decrease in synaptic sensitivity (caused by a flood of neurotransmitter, meth) and a steep drop in neurotransmitter activity once the meth is gone.

Stimulants shred the cells' mitochondria. Psychosis is a metabolic problem.

>Each of these dimensions was predicted by unique abnormalities of brain networks,” Satterthwaire explained. “However, all dimensions were marked by abnormally high levels of connectivity between the default mode network and fronto-parietal network, two brain regions that usually become increasingly distinct as the brain matures. This loss of normal brain network segregation supports the hypothesis that many psychiatric illnesses may be disorders of brain development.”

I wonder if that points to failures of the synaptic pruning that's supposed to happen in preschool-to-primary-school age kids.

Meditation and possible some states of enlightenment(fully quiet mind), are about reducing activity in the default-mode-network(DMN). Psychologically this is seen as a reduction/loss of the sense of self(or more exactly stopping self-referential thoughts)

On the other hand, a common association[1] with several mental illnesses, is rumination("repetitive, self-directed thought") - which actually fits the current study. One wonders if specific mental illness cause rumination and why ? or is it backwards ?

[1]https://www.psypost.org/2016/07/study-suggests-rumination-tr...

Sure, but synaptic pruning happens at all ages. After all that's how we learn anything and forget unimportant experiences. It would be interesting if one could strengthen certain interconntions while weakening the mental illness causing ones.

The other day I was looking at Computer Science courses offered the Georgia Institute of Technology, and "Behavior Imaging" sounded interesting. It's part of the realm of computational behavioral science

> Course Info: http://rehg.org/teaching/introduction-to-behavioral-imaging/

> Computational Behavioral Science: http://www.cbs.gatech.edu/

This looks interesting, anyone taking/has taken the course?

Reply on news.ycombinator.com