haldujai

14 ms·

151.
by haldujai 3y ago
> On average, 33 cents of every insurance dollar in the US goes to paperwork and paid negotiators on the doctor or insurance company side. That's misleading, this is the percentage quoted for overall non-clinical costs. The better n
152.
by haldujai 3y ago
Needing a referral (an assessment by a physician that something is medically appropriate) is not the same as needing preauthorization (insurance company agreeing to pay) but in any case hence the following "you have to go through your
153.
by haldujai 3y ago
> I’ve been in a Kaiser plan most of my life and the “rigidity” doesn’t bother me, maybe I’m just used to it. I should have said it can be* very annoying, the rigidity isn't an issue for everyone and it really depends on your situat
154.
by haldujai 3y ago
> It has been around for decade. Assume that's a typo and you mean decades. Kaiser dates back to the interwar period and the modern version of it to the 70s/80s. > For the majority it seems like it's a cheap but very ri
155.
by haldujai 3y ago
I think the commenter means "you can have preferred access". PPO = Preferred Provider Organization. You have a better deductible/coinsurance/copayment for "in-network" (preferred) providers who have negotiated
156.
by haldujai 3y ago
I don't understand, this paper is describing how EBM works today and calling it EBM+ for some inexplicable reason. This is a long winded rant basically concluding with face validity is important...
157.
by haldujai 3y ago
> Chinchilla scaling was only good for academics I don't know if it's only good of academics, the point as the paper (as it says) is a scaling law for optimal loss given a fixed compute budget. By design it doesn't address
158.
by haldujai 3y ago
The discussion at the end of this article starts to get to the problem with extrapolating. Llama1-65b (roughly Chinchilla optimal) and Llama2-34b used similar compute and although Llama2 not directly comparable is the closest comparator wit
159.
by haldujai 3y ago
> data you've only seen once Is this still true given that they're upsampling in the pretraining dataset? I don't recall any details on how and to what extent they did this in the Llama2 paper but presumably some fraction
160.
by haldujai 3y ago
You're (correctly) addressing some of the issues in biomedical research but even the most rigorous and validated science does not necessarily translate into clinical outcomes for a laundry list of reasons. Evidence based medicine inclu
161.
by haldujai 3y ago
> You lost me here - because the science based one is just stating unproven statements to come to some conclusion. If you were to expand on why those two sentences are "proven" you'd likely come to evidence based reasons.
162.
by haldujai 3y ago
A simple example is (using the traditional understanding of these differences): Science-based: Cholesterol levels are associated with major adverse cardiovascular events (MACE), statins lower cholesterol levels and therefore statin therapy
163.
by haldujai 3y ago
I would add that EBM took over because science-based medicine (at least the historical definition, how it's used now is more of a hybrid) didn't work out too well. Often the mechanism thought to be understood is incorrect or more
164.
by haldujai 3y ago
> The quality of studies in systematic inputs is low and even Cochrane's reviews have been critiqued for missing things. Criticism is good, expecting perfection is not. Quality issues are a problem of the underlying research rather
165.
by haldujai 3y ago
Bad example on my part in that case. I thought some art is copyrighted or am I mistaken? If so replace Mickey Mouse with something copyrighted
166.
by haldujai 3y ago
1. If you know your copyrighted material was in the training dataset is that not sufficient? 2. From a legal perspective do you actually have to prove it's embedded in the gradients? If I draw an exact copy of Mickey Mouse from memory
167.
by haldujai 3y ago
I wonder if that analogy represents the same thing. Speaking purely from a non-legal perspective on the ethics in my mind: When you use Photoshop on propriety data you're providing the original data and choosing what manipulation to ma
168.
by haldujai 3y ago
Interesting. Probably did shift then, it would fit the pattern of epidemiological changes taking a while to percolate to radiology and as it's far more likely we miss crypto on MRI than toxo we probably wouldn't notice a change in
169.
by haldujai 3y ago
Just in case it's not clear I am by no means claiming domain expertise, merely stating that what I was taught and my understanding was similar to the initial comment I replied to hence the caveats and soft language. My statement should
170.
by haldujai 3y ago
Yes. Microbes once in the blood stream can cross/bypass the barrier by a few ways such as by crossing endothelial cells, crossing loosened or disrupted tight junctions or by entering a host cell that then traverses the barrier. Normall
171.
by haldujai 3y ago
It depends on what type of infection we are worried about and what structures are involved but generally an MRI and a lumbar puncture for cerebrospinal fluid analysis to start. If there's an abscess a neurosurgeon can stick a needle&#x
172.
by haldujai 3y ago
I think the commenter is referring to acquired HIV encephalitis caused by direct HIV infection which I was also under the impression was the #1 CNS infection in HIV patients. My teaching (in radiology) was HIV > toxoplasmosis > crypto
173.
by haldujai 3y ago
The pump isn't the hard part.
174.
by haldujai 3y ago
There are faster acting insulins creatively named "faster aspart" and "ultra rapid lispro" with the latter beginning to appear in the bloodstream at 2 minutes and reaching 50% effect around 20 minutes. With that said the
175.
by haldujai 3y ago
You seem to be misunderstanding how these devices work. Bihormonal pumps do not mean continuous infusions of both insulin and glucagon. The pumps pulse insulin when you're high and glucagon when you're low. They're not both a
176.
by haldujai 3y ago
> There is a what seems to be a significant number of people who don't "feel their lows." Severe hypoglycemia to the point of what was described (death) is not reported in any of the recent device studies. Level 2 or moder
177.
by haldujai 3y ago
I think what you're missing for this to make sense is what is the desired outcome. For type 1 diabetics there are three important ones: 1. Time in severe hypoglycemia - ideally 0% 2. Time in severe hyperglycemia/diabetic ketoacido
178.
by haldujai 3y ago
> Agree that patient input shouldn't be necessary, but to replace it we will need to include other inputs besides CGM in a systematic way to get the optimal results. I'm not going as far as to claim Medtronic's approach (I
179.
by haldujai 3y ago
Inreda integrates the CGM, [1] has more details on the setup. > I'm very familiar with one of the most popular, closed-loop system combinations in the USA and I definitely don't feel the hypoglycemia problem is anywhere near &q
180.
by haldujai 3y ago
> However, if your brakes (glucagon) can sometimes fail completely, that could cause you to die almost immediately Failure detection is via alarms to trigger patient action based on the continuous glucose monitor (which has a different s

More ›