Physician Matt Zirwas proposes using cheap, frequent ultrasound scans to monitor cancer anomalies instead of ordering immediate biopsies

Very good post on how things might change with largescale scanning done regularly. It's a possibility, and data will tell.

@thirdpairdemon @VictorTaelin You live in a world with broadly superhuman intelligences. Heavy lifts are becoming light at a shocking pace.

@VictorTaelin My concern is having enough capacity to read all of the scans.

Going from 100 scans a day where 20 find "something" and 2 are cancer to 1000 scans a day, 50 find something and 3 are cancer, but you missed one cancer because you were too overloaded to read the scans properly.

@MaxLittle92 good, you can leave that to the ML researchers and software engineers

I mean yes but this is for well established methods with pretty high resolution, you don't really need to be a doctor to compare this to their early results, and ultrasound scanning of mammary gland is very noisy anyway so this is the best we have. I'm not sure though what's the sota on ML driven detection, likely much better but that's still far from clinical integration

@VictorTaelin Ultrasound is super noisy in general and as anecdote I can tell you that only 1% of doctors doing ultrasound scanning are competent and reliable, others just straight up random guessing

@VictorTaelin I really appreciate people who can take a step back and reevaluate their thoughts, if more ppl were like that world would be a much better place

Bro trust my I'm MD (for real). Sensitivity/specificity rate of mammary gland CT scanning is so bad that for one woman with real diagnosed tumor there are 25-60 women with false positive findings, and some false negative cases as well. For thyroid gland screening it's widely known that it doesn't help with anything at all on average so as a program it doesn't make sense, for individuals well depending on how anxious they are.

@VictorTaelin most doctors are missing the key fact that it's an ultrasound machine, ie mechanical wave and not radiation!

@GrigoryEvko if the data is bad then it is noise and a competent system just ignores it, that's the whole fucking point

stress

@GrigoryEvko if it is all noise and zero signal it will just be dropped my dear

@VictorTaelin So all high profile surgeons keep close friendship with radiologists they trust and sometimes most established radiologists are noise generators as well and really accurate ones are getting bullied usually

@VictorTaelin sadly it’s always been hard. people rarely change their opinions, even when confronted with incontrovertible facts. perfect example are “intellectuals” who could not handle that communism led to millions died in the gulag. 22min in…

https://youtu.be/vw6QaYB4Obk?is=b3KWpKHjUJQeu58D

@VictorTaelin @GrigoryEvko Then why do the scan if you’re just going to ignore it

@VictorTaelin All these false positives posts lead me to believe that the medical industry has zero fucking clue about anything and have gone unchecked

@VictorTaelin It still assumes that the disclosure of the information doesn’t lead down one of the bad paths that wouldn’t have happened otherwise

@GrigoryEvko @VictorTaelin can you sort of see how "make a better ultrasound scanner and protocol for using it" is not answered by "ultrasound sucks right now"

@VictorTaelin “If the Midjourney ultrasound is high resolution, harmless, inexpensive and convenient”

these assumptions are doing a LOT of heavy lifting

@VictorTaelin Is the a-ha moment realizing techies are stumbling over themselves for unproven tech because Elon commented positively on it?

@VictorTaelin So population wide we keep 25-60 women anxious (depending on the qualifications of radiologists in the region) for 1 woman we can potentially help, I still don't know how to internalize this