Post

Conversation

Blake Byers
@byersblake
You'll see a lot of doctors come out "against" this kind of broad screening system. They can even get quite agitated about it. This resistance stems from a well-established clinical consensus: traditional population-level imaging fails to improve health outcomes because false positives and invasive follow-ups do more harm than good. But this view suffers from an obvious blind spot. Existing studies rely on static data and completely ignore time-series imaging. And time-series is ignored because we haven't been able to afford to do high frequency imaging at population scale. Clearly, time series is going to be immensely more valuable than a single image. If you drop costs, value can go from 0 -> 1. On a more fundamental level, the argument against screening rests on an obviously false precept "More information is bad" -- just clearly untrue. More information better, you just have to interpret it correctly.
Quote
Midjourney
@midjourney
Announcing a new division of Midjourney called "Midjourney Medical"
The media could not be played.
29.7K
Views
View quotes
David Watson 🥑
Post your reply

Arturo LoAIza-Bonilla, MD MSEd
@DrArturoAI
Nah. I do Prenuvo MRIs for myself and family. The "resistance" is because ultrasound waves do not penetrate bone and scatter with air. Not a good screening system if physics do not allow it. No matter how many tests you do would mitigate this unique factor.
1.2K
Blake Byers
@byersblake
Im not sure what
@midjourney
is trying to detect, but I suspect its not meant to be a full replacement for MRI. I could see some subset of things that benefit from higher frequency scans while being stuck at larger voxel sizes. Also, MRI is particularly bad at moving tissues
1.1K
Misha G.
@tastybits
There is also a misunderstanding that the false positive rate we have baked into historical studies is driven by human error. That rate is going to be lower with abundance of intelligence. This is anecdotal, but for what it’s worth:
Quote
Misha G.
@tastybits
This take is so strange to me. Inexpensive whole body scans are a game changer. I did a gratuitous ultrasound last year. No reason for it. Zero symptoms. Just thought it would be a good idea. Scan showed a 4.5cm tumor on my left kidney. Miraculous robotic removal at Baylor x.com/septisum/statu…
435
Alphonso Dente
@AlphonsoDente
Gibberish… take a real world example: pt has had 200 magic US water baths. Around the 150th new blob started being noticed in the right lobe of the liver. Growing very slowly. Magic water bath AI says probably nothing. Pt goes home can’t sleep at night , sometimes I get pains on
39
M. U. Dalmis
@iandcosmos
Right, but it will still take many years of data collection and research to understand the medical impact of such findings and what to do about those findings. So in the short term I don’t expect much medical impact, but in the long term the collected data might open new doors.
350
Paul Bohm
@paulbohm
I’ve had so many discussion with medical “experts” who try as hard as they can to minimize information
76
Abe Murray
@abemurray
This 1000% They cannot imagine a world where you build the better data set, then do something better with it And they paternalistically don't trust us to use this data well
104
Jason Jin
@jasonljin
have so many angry doctors in my replies today in response to me saying: - cancer is bad - we should better detect it early - problem of false positives are largely due to our (currently) limited episodic + expensive diagnostic methods
Quote
Jason Jin
@jasonljin
Replying to @JOSEPH45075332
how do you know it’s a false positive? the point is you don’t, not from one scan but today, the data is incredibly sparse and expensive to collect. you are then presented with a binary decision when you have a false positive: ignore it -> too risky escalate it -> expensive /
342
Liping Bai 白莉萍
@bai_liping
Fully agree. Being able to see how things evolve is more important than a highly accurate snapshot. That is the same argument Obama used to to repudiate comment from his Paster. While that Paster might had an accurate snapshot view, he failed to see how things has changed.
52
Hüseyin S. Bozkurt,MD,AGAF
@HSEYNSANCARBOZ1
This is first step of “scan and treat” model
Quote
Hüseyin S. Bozkurt,MD,AGAF
@HSEYNSANCARBOZ1
Replying to @jasonwilliamsmd
This is also the beginning of a path leading to the "scan and treat" model. x.com/hseynsancarboz…
666
Bubble
@Tripleboulders
If there's a 1% of false positive, then just do the imaging a few times and it would be 0.00000001%.
470
Jarrel Seah
@jarrelscy
Radiologist here who actually thinks this could be good. But care should be taken - more information is good if you correctly calibrate your Bayes factor. Many patients ask why their MRI found something their CT did not. Patients receiving USCT needs to know what it can’t do.
246
Kevin Morrill
@MisterMorrill
Disagree. Had my gall bladder imaged multiple over an 18 month span, each time revealing what appeared to be growing number and size of polyps. Albeit with weaker tech. They removed the organ only to find it was an aberration.
230
Dr. Unk
@CRA1pf
Not against it, just haven’t seen any proof the “images” actually qualify as diagnostic
294
George Darroch
@georgedarroch
notice that very few doctors oppose looking at the skin for potential melanomas, annually or more frequently if necessary. they're unable to extend that thinking to inside the body.
Quote
George Darroch
@georgedarroch
Replying to @rheum_ai @JohnAnthonyMD and @hankgreen
one of the more interesting comparisons is with skin. a trained prof (or AI with imaging) can view the data (skin) with no indications and follow-up is easy. biopsies are lower-invasiveness, and removal is usually low-complication if we were able to apply this internally..
190
Araz
@Araz_io
Time series wouldn’t make much difference. Both malign and benign lesions grow over time. Not all malign lesions will cause mortality/morbidity during you life time, the investigation will likely cause more damage than good.
344
Hecateus 🇩🇪🇪🇺🇺🇦
@Hecateon
Nah, their image quality is bad and the process inapplicable to many patients. This'll be a gimmick for healthy young people unless it fundamentally changes.
30
Bla Blu
@BlaBlu6969
But they will give standing ovation to a new big pharma protocol that extends 0.01% cancer patients lives by 2 months.
216
Silva, J.P.
@ertwro
“Longitudinal > cross-sectional” is introductory epi/stats, not a revelation. Serial data helps trajectories but does not eliminate overdiagnosis of indolent findings, lead-time bias, or cascade decades of RCTs on imaging screening prove this. Mr Value 0 -> 1
33
sdmat
@sdmat123
"If we had imaging results we might have to treat you, and that's bad" is just fundamentally irrational. Take a Bayesian view of diagnosis treatment and there it is all upside. The problem is the medical profession finds it convenient not to have to make the judgement calls
65
𝙳𝚊𝚟𝚒𝚍 𝙲𝚊𝚗𝚎𝚜
@CanesDavid
Blake listen… this time-series theory is a complete fantasy because it misunderstands both human biology and basic medical ethics. First… tumors do not grow like clockwork in a predictable, linear timeline. An aggressive, lethal cancer can go from completely invisible to
17
Carl Lumma
@clumma
Exactly
Quote
Carl Lumma
@clumma
What I call "practicing medicine" doctors call "longitudinal monitoring"... and never do it.
203
T Ay.
@ayedtay
Yes sir
Quote
T Ay.
@ayedtay
Yea I don’t know anything about medical imaging but there’s that fake experts gatekeeping vibe that tells me no weight should be put on so-called experts dismissing them These people will keep repeating it’s impossible and a fraud right until they actually do it Go x.com/doodlestein/st…
45
Unix
@Unixsystem13
Importantly, AI is pretty damn good at identifying worrying patterns out of huge piles of data. It doesn't even have to be particularly great, just eliminate the obviously-healthy from the maybe-unhealthy and send the rest in for followups.
49