I'm an ER physician. I have worked in a poison control center. I take care of the people that drink IPA. I work on the Navajo Native American reservation, where this isn't terribly uncommon.
795 karma · joined May 5, 2014
F21A6194C9DB9899CD09E24E434B2C14B8C33422
[ my public key: https://keybase.io/n8henrie; my proof: https://keybase.io/n8henrie/sigs/WoQvsZKguJVoMfO7Fxk2mnWhN1Xh0cLZuDGQnxpwhWY ]
I'm an ER physician. I have worked in a poison control center. I take care of the people that drink IPA. I work on the Navajo Native American reservation, where this isn't terribly uncommon.
It would be asinine to study the effects of parachutes on survivability of jumping from airplanes, hypothesizing that they would help, but conclude that the "conjecture must be wrong" because the sample size was 2 and it failed to reach statistical significance, or because the airplane was on the ground.
Would you feel differently if the study period was only 1 day instead of 25?
Or maybe 1 hour?
Would it then be reasonable for them to speculate that the methodology might contribute to the failure to reject the null hypothesis?
Reviewing my work, only needed an hourly aggregation, which was similarly easy in polars and pandas (I misspoke about being easier) -- what I found easier was grouping by time data that wasn't amenable to `resample`.
In polars I had no problems using a regular group_by with a pl.col.dt object, whereas in pandas I remember struggling to do so, even though it seemed straightforward.
Sorry, I wish I could remember more details; this was probably 5 years ago that I was writing the pandas code and just converted it to polars about a year ago, so it's possible that I just got better at python in the meantime (though I was writing much more python back then). And of course a rewrite is likely to feel easier the second time.
The other confounding issue is that the eager pandas code crashed with OOM regularly and took several minutes to run, whereas polars handles it very well (which I'm sure to some degree is it optimizing things that I could have done manually), but this made iterating on this codebase feel much less onerous.
I've had trouble determining whether one timestamp falls between two others across tens of thousands of rows (with the polars team suggesting I use a massive cross product and filter -- which worked but excludes the memory requirement), whereas in pandas I was able to sort the timestamps and thereby only need to compare against the preceding / following few based on the index of the last match.
The other issue I've had with resampling is with polars automatically dropping time periods with zero events, giving me a null instead of zero for the count of events in certain time periods (which then gets dropped from aggregations). This has caught me a few times.
But other than that I've had good luck.
Unfortunately sandbox-exec isn't really documented (and supposedly deprecated?) so trying to sort this out is a bit of a headache.
As a physician with a degree in nutrition, I find most HN threads about medicine, nutrition, or the health system very frustrating, and in spite of it being the only area in which I'm formally qualified to opine, it's also where I've earned most of my downvotes (which prior to 500 internet points kind of mattered). I eventually learned to just bite my tongue here.
Curiously, I think my "hacker spirit" is what drove me to medicine. In undergrad, I was spending 5+ hours per day reading medical literature (mostly nutrition, endocrinology, exercise science) and had decided that doctors were mostly idiots, and eventually decided to change majors from ME to pre-med.
In medical school, I was a real handful to many of the lecturers, which I kind of regret now. Ends up it is really easy to publicly humiliate someone when they mis-state or misunderstand a minor detail that you've studied in depth, even if they have far greater expertise in the field.
In medical school, residency, and practice (EM), I've learned a lot about why things are the way they are. In my field, many patients are entirely obsessed about some problem and completely lose track of reality when discussing it. It becomes part of their identity. They don't know how to read or evaluate medical literature, and they lack the background knowledge to the extent that even beginning a discussion is onerous.
It's like your uncle who is far to one side or the other of the political spectrum and is well read but only in support of his biases -- yes, you might learn something this time, but do you really want to get into long discussion to find out? He already "knows" all the answers to many unstudied questions, and knows why the studies are wrong for the others. There are so many of these patient encounters that it is entirely infeasible to engage with even a small proportion of them and still get your job done, so you learn to nod, smile, and move on to determining whether or not a life-threatening emergency exists or not. You definitely don't have time -- at least in my field -- to really listen and consider all of these. Unfortunately it's not always easy to differentiate symptoms that are possibly psychological, exaggerated, self-limiting, or impractical / impossible to diagnose in my clinical context, from those that present opportunities for me to make a difference even if not a true emergency.
I still think that many doctors are idiots that lack critical thinking skills and self-reflection, and unfortunately almost nobody outside of academics reads primarily literature that isn't forced on them (MOC). I've tried to show many friends and colleagues how to set up RSS feeds for pubmed queries for their topics of interest -- zero people have seemed impressed. In contrast, I think much of HN leans heavily on representativeness heuristic and doesn't consider likelihood ratios when evaluating their test results.
Anyway, thanks for the considerate comment.
Not entirely unreasonable to consider, but if they need defibrillation and are unconscious / unable to roll themselves over, they might have bigger problems. Just the movement of properly executed CPR is going to be considerable.
If anyone else is curious about the "long-term" part
Currently running in docker, which was a godsend when I migrated from Raspbian to NixOS about 18 months or so ago with no issues.
Have used an SSD wherever possible, perhaps that's why it's been so stable.
I use the GPIO, which is one reason I've stuck with the Pi!
https://n8henrie.com/2023/06/write-a-firefox-extension-in-py...