ELIZA wasn't the only chatbot in the late 1960s... this JAMA+AI podcast explores how their creators took very different paths in beliefs about AI in medicine
Not to take anything away from the meta-analysis and the important points in the blog post, but there's a glaring error in the abstract: "the weighted average effect sizes were as follows: social isolation odds ratio (OR) = 1.29, loneliness OR = 1.26, and living alone OR = 1.32, corresponding to an average of 29%, 26%, and 32% increased likelihood of mortality, respectively."
The authors mean increased /odds/, not likelihood (probability). WHy does it matter? Well, when your whole paper is a statistical exercise, misusing basic statistical language in the abstract is not a great sign.
sorry, intended to elaborate: When I query my health system's deidentified research patient data registry, there are multiple individuals with both a diagnosis of schizophrenia and a diagnosis of congenital (cortical) blindness. One can of course quibble that claims codes are unreliable (ie, these people really have different diagnoses), but my real point is that the entire article is based on a very limited analysis that could readily be refuted with any of the claims/health registry data sets....
also amusing to see this downvoted, people on HN don't like to see their pet theories questioned.
>"I feel like perpetuating this mindset is the ultimate sin of modern psychiatry and the entire industry of psychiatric pharmaceuticals"
Hmm. Modern psychiatry is usually accused of the opposite.
Also interesting that leading comment on an article about how many (non-CNS) meds have CNS effects becomes yet another opportunity to criticize antidepressants. HN community's hostility to treating brain diseases (while emphatically supporting people with brain diseases) is perplexing.
Ironic that folks who work in CNS drug discovery spend much of their time worrying about how to get our molecules /into/ the brain... and less energy thinking about keeping non-CNS meds out. For what it's worth, though, these 'off-target' CNS effects have also prompted a lot of enthusiasm for repurposing existing FDA-approved drugs for new indications, since their safety is already established.
the studies (which admittedly are short-term) reflect remission - i.e., resolution of symptoms. This is a critical point: the goal is resolution of symptoms, not partial improvement.
The closest thing to a cure (for some people) can be cognitive-behavioral therapy, or long-term antidepressant treatment when indicated.
for moderate to severe depression, the kind most posters are talking about, they're consistently better than placebo. It is true that placebo is often effective too. But most people I talk to would rather have that extra chance of getting better.
actually that statement is itself rather presumptuous. How would you know that those of us who repeat it don't know? I can state with authority that may of us do.
A colleague of mine calls suicide 'a permanent solution to a temporary problem' - not to diminish it, but to recognize that for most people the desire to die is a temporary state.
Whenever depression comes up on HN (as it does with surprising frequency), I'm both touched by people's willingness to share their own stories, and frustrated by otherwise very rational and logical people's speed to dismiss data. Initial treatment (meds or evidence-based therapies) work for about 1/3 of people; subsequent treatments work for another 1/3; and there are up to 1/3 where multiple treatment trials fail them. (This comes from STAR*D, plus Cipriani Lancet meta-analysis, plus vast CBT literature). But, there are a number of next-step treatments showing promise (rTMS, esketamine, variants of CBT, and so forth).
"Obscure Newspaper Discovers 40-year-old Concept Widely Used in Medical Literature"
The concept of QALY goes back to 1976[0] and is a mainstay of cost-effectiveness literature. Fair to say there has sometimes been resistance to its application in the US, where we like to pretend that we are equitable in providing care. But odd that WSJ seems to have just discovered this idea.
Fun lithium fact: while most people start the lithium story with Cade (indeed, there's a form of lithium-responsive bipolar disorder which was sometimes called Cade's disease), it was previously used by Lincoln's surgeon general, William Hammond, to treat manic excitement (in the form of lithium bromide). But Hammond thought it was the bromide, not the lithium, that did the trick.
please be careful in equating the two: there are decades of studies, including large trials, demonstrating benefit for lithium. It's considered gold standard treatment for bipolar disorder even in an era where big pharma pushes hard for newer and more costly meds.
there is modest and inconsistent evidence from small trials that chromium picolinate, a dietary supplement, may be beneficial in depression.
Oh, and note that chromium is more expensive than lithium too!
Here's my use case: as a scientist, I'm often in the position of needing to review multiple 20+ page grants or manuscripts, resulting in hundreds of pages at a time. I find it far easier to read closely and retain info when I can write directly on the documents, then circle back to collect my comments, vs typing directly. At a certain point, looking at reams of paper coming out of the printer made me feel slightly sick to my stomach. Plus, I appreciate that I'm not left with those stacks of paper sitting on my desk.
(Before someone does the inevitable HN calculation about why the materials used to make the Remarkable are far worse for the planet than all the trees killed to print grants - yes, you're probably right. YMMV.)
I had tried the various ipad and equivalents, and the RM's big advantage is that it feels like reading/writing on paper, mostly - something about the matte screen and stylus, though the stylus is sort of craptastic.
As others note, the software is clunky with annotations moving around on the page, and there's a bit of lag. But, I have yet to find anything better. I do wish there were an easy way to collect all the annotations in one place, but my handwriting as an MD is appallingly bad so that may be an impossible task.
I'd be curious to hear you define biologically valid. The heritability (variance explained by inherited features) for those diagnoses ranges from 0.3-0.8 - greater than many other medical disorders - and for many of them there are identified genetic variants. Maybe that's some of the 'exciting research' you allude to?
I agree that they are used as heuristics - and indeed were developed to be heuristics, dating back to the RDC.
Before everyone goes bananas citing Goodhart's law: many universities and academic medical centers in the US don't care at all about impact factor - they care about grant $$, period full stop. (They appreciate the occasional high-impact paper that they can use in marketing materials, but it's really all about the $$.)
And for what it's worth, I've almost never heard impact factors discussed at NIH study sections, where investigator quality is explicitly on the agenda. Reviewers talk about relevant prior publications in the field, esp in marquee journals. [this latter feature is the reason we don't just put everything on biorxiv or equivalent and move on.]
For everyone complaining about peer review, another alternative to open review is paid reviewers who focus on a particular topic. For example, I'm an editor at a journal where statistical reviewers are paid and evaluated, while the remaining (content) reviewers are not. This ensures that we have a rigorous review of the statistical methods, at least. I do think payment increases the reviewers' responsibility and conscientiousness, and the overall quality of the subsequent articles. (nb This probably has been tested somewhere.)
BUT - paid reviews are not feasible for many journals, especially open access/low-cost journals where margins may be thin. (Conversely, Elsevier should be able to pay their reviewers in gold bullion, rather than taking advantage of the scientific community's altruism, but that's another conversation).
Those are all very real problems (and as you point out, they impact grant review as well). On the other hand, fully transparent peer review doesn't necessary address most of them. Reviewers who have to sign their reviews may be reluctant to anger colleagues (or try to curry favor with them).
As you presumably know a many journals have experimented with open peer review, but editors still need to police the reviews to look for bias. It solves some problems but creates others.
nope (re being sure their ability to handle health benefits has improved)... just used them for a startup and their response to questions about integration with health benefits was essentially (3 days later) - um, we don't do that, but we'd be happy to sell your name to a broker.
agree the interface is pretty, but failure to really support benefits will be a dealbreaker for many startup/small business folks.
Psychiatrist here... bravo. The feature creep, lack of rigor, and overtly commercial nature of most CBT lite apps is precisely the reason I've been reluctant to encourage their use, vs carrying around an index card. Do one thing and do it well.