joshgel

24 ms·

211.
by joshgel 8y ago
>it learned how to play chess well This. Learning to play a game is one thing. Learning how to teach computers to learn a game is another thing. Yes chess programs have been good before, but that's missing the point a little bit. T
212.
by joshgel 8y ago
Salt is one of the bodies most hormonally regulated substances after maybe glucose. The electrolytes Na and Cl are routinely and easily measured on basic blood chemistry panels. My patients with migraines don't have low salt. cant say
213.
by joshgel 8y ago
Doctors use this model for evaluating/trying to prevent medical errors as well
214.
by joshgel 8y ago
I loved this book. The destruction a few thousand remotely controlled cars could wreck is terrifying.
215.
by joshgel 8y ago
Isn't this what Trump did too?! Refused to blink despite all the 'hits'...
216.
by joshgel 9y ago
Ya, for all the faults of social media, and we know there are many, the ability to publicly shame companies doing sketchy things into changing is a plus. Who knows how this falls in the balance between good and evil?
217.
by joshgel 9y ago
Physician here: CAR-T is a potentially amazing technology, but really only works in a small sub-set of the patients for which it is tried. For now
218.
by joshgel 9y ago
Wow this is super helpful. Especially for people who approach ML from a coding perspective and don't have as much math background.
219.
by joshgel 9y ago
wait time == infinity, if you can't get access to care you need at all!
220.
by joshgel 9y ago
But you are underestimating the degree of variability between providers. Some providers may have felt the XRay was necessary and others may have not. In general, I find that patients have a sense that there is always a 'right' ans
221.
by joshgel 9y ago
As a physician, one of the things not touched on in this discussion and only briefly mentioned in this article is the harm associated with hospitalization. For many of my patients discharge planning is one of the toughest conversations to h
222.
by joshgel 9y ago
I did both and found fast.ai so much easier to understand for someone without a background in math, like me.
223.
by joshgel 9y ago
I would highly recommend starting with course.fast.ai. Really helped me.
224.
Predicting Cardiovascular Risk Factors from Retinal Fundus Photographs Using CNN(arxiv.org)
2 points by joshgel 9y ago|0 comments
225.
by joshgel 9y ago
So ya that's what all cardiologists thought. Many still think this. They may even be right. But we don't have EVIDENCE that this does anything to prevent death, heart attacks or even reduce chest pain. To be clear though, they are
226.
by joshgel 9y ago
Loads. A functional EHR (UX). An easy way to keep track of what medications a patient is one and has been on (?blockchain). Better quality and cost measures (big data).
227.
by joshgel 9y ago
Prove it. In a randomized, placebo-controlled, double-blinded trial. There are plenty of quality and health savings organizations that would fund such a trial if there were any hint of benefit. But there isn't. I'm glad to hear th
228.
by joshgel 9y ago
Sure. Or those hospitals gamed the system and didn't admit patients who should have been admitted. Or plenty of other explanations... We call readmissions rate a Surrogate Marker. Something that isn't a hard outcome that we care a
229.
by joshgel 9y ago
Agree. My dad, also a physician, blames himself for being overweight. But if "everyone" is overweight, can it really be their fault? Surely "the system" has as much or more blame to share.
230.
by joshgel 9y ago
Before you become "so dismissive" of traditional physicians, you try getting someone to radically alter their lifestyle for the promise of potential benefit in 10-20 years. Even if you can point to a few people that have done it,
231.
by joshgel 9y ago
Interesting you bring up heart failure readmissions. Not saying they aren't a problem, but here is a fascinating study in JAMA that argues that places with lower readmissions have higher (worse) mortality. https://jamanetwor
232.
by joshgel 9y ago
I'm sure we heard this frequently in the early days of the public internet. "What are we going to use this for?" Not a lot survived those early days, but enough did.
233.
by joshgel 9y ago
While I loved this post, could have also been helpful to include more info about Pythagorean Means (geometric mean and harmonic mean). Worth looking into depending on the type of variability seen in your data. EDIT: just saw this was mentio
234.
by joshgel 9y ago
"yet" +++
235.
by joshgel 9y ago
This is exactly how science works. They report this possible case. Maybe there is never another association and the issue is forgotten, because there is no causal link. But maybe, someone else, some place else, who never read this paper, no
236.
by joshgel 9y ago
These are standard doses of all these medications. Its the dose of tylenol often used in the hospital and wasn't dosed every 2 hours it was dosed once. For the opiates we either give 1 or 2 pills, here they gave 1. Perfectly reasonable
237.
by joshgel 9y ago
This study only looked at acute pain. Generally those medications are safe for acute pain. (Unless by 'safe and effective' you mean NEVER have side effects, in which case we probably don't have any 'safe and effective&#x
238.
by joshgel 9y ago
fascinating book! love the follow up too!
239.
by joshgel 9y ago
I'm reading the book Sapiens right now, which I highly recommend. Lots of interesting facts. But also a broad explanation of the theory of human development. Most interesting fact I learned: At one point there were as many as 6 human s
240.
by joshgel 9y ago
I'm early 30s and relatively tech savvy, the snap app is the hard app to figure out.

More ›