Lan Dao, Women's Health Fund founder, argues cheap AI scans solve diagnostic false positives through frequent monitoring

Yep, most of these things are benign or resolve themselves without intervention. That's why I say the barrier is cheap monitoring. Getting multiple MRIs to keep an eye on an anomaly is unreasonable. You have to treat everything you see as a potential problem, because no one is going to schedule and pay for a sequence of imaging.

One full body scan a year? Useless. One a week? Different story.

That much data might change our understanding of the body's fluctuations, also, which means those false positives might reduce, and we'll know 'this is typical of most humans, this is when it becomes actionable.' But we just don't have the data to have those insights. Yet.

I asked an oncology surgeon friend, maybe 10 years ago, "Why don't we just get a full MRI as part of the yearly physical?"

1) Cost. 2) Apparently, tumors and cysts and such appear - and disappear - all the time, and they don't do much of anything. We would end up doing a lot of unnecessary surgeries.

(that's my interpretation from hazy memory, at least)

I think this is an important point.

False positives from full body scans are a problem because we can't handle them very well. We could just get better at that, too.

Cheap repeatable scans would likely be step 1 in the process of improving false positive handling.

Yep, I've pointed out the same thing in other comments. If scanning is safe and cheap, its easy to avoid invasive biopsies in favor of just another scan in x days.

Also if we have lots more scanning, we have lots more data to train AI in image analysis with much more accurate differentiation between noise and real concerns.

The more pushback, the more bullish I'm going to be as effective disruption always gets pushback from the entrenched technology.

@ryanssking1 @mcahogarth The transmissible diseases inside a hospital that can be carried in the water need an exhaustive cleaning after every use to make the device safe. You can't dump patient's data. That's a HIPPA violation. Processors need maintenance, space, and ventilation. Costs can't go down

We are still early.

@mcahogarth It needs 21 servers, millions of nano transducers, creates 1TB images, and every drop of fluid would have to be disposed of to avoid health hazards. The cost will be HIGHER

@AutumnStranded For now. Those are solvable problems. I can think of several solutions off the top of my head, and I'm not even an engineer or specialist.

@mcahogarth Pathologist here You are confusing monitoring and diagnosing You can perform a 1000 scan as long as you dont put it under my microscope you will not know what it is. Growth is different from malignity and a cancer can be Tiny, spread, kill and dissapear

@mcahogarth @DoggieLine we'd have to completely change the culture of panic and figure out how to balance between waiting to see what happens vs jumping straight to needing to take surgical action asap. I think we're going to be surprised how often human bodies just take care of problems.

This is my take, also. A lot of things are gated behind lack of data.

I think of self-driving. Self-driving was "obviously" not possible until it was, because you needed enough training data to reach the tipping point.

I appreciate the people who are speaking from positions of experience, but experience becomes obsolete over time.

The water is small, it’s less water than the average person uses in a shower, it’s literally taking a bath. The servers are for processing, they run a job and move on. We need to know the jobs per hour it can run. 1TB of storage is $15-25 if using SMR. Once processed, we can dump the data. The nano transducers are part of the capex of the machine. So what, even if it’s $500 per job. People will pay it. Over time. We’ll improve data compression and server performance. The cost will continue to fall.

@mcahogarth Brain, lungs, gut. Can't image those. And other spots where the tomogram reconstruction suffers from data gaps due to attenuation of the ultrasound waves by those structures. Not to mention standing and immersion aren't always an option for the sick.

@catphonics @DoggieLine Me too, and I'm interested in how this will change how we approach medicine. We need a paradigm change.

@mcahogarth Ok but we can already do very cheap and very quick better whole body screenings anyways (MRI). If one wants apples to apples, that is (ie not diagnostic quality scans which are longer and have multiple protocols)

@mcahogarth They said the same thing about cheap personal DNA testing, and they were wrong.

Personal DNA testing has altered lives for the better. Doctors just don’t like it because they don’t like informed patients that ask questions.

@mcahogarth Exactly

@mcahogarth You need pathology flags for AI to improve. How do you think we deliver that data? And btw. Midjourney does not use AI here. It's just hardware and software.

@mcahogarth

@mcahogarth You can be scanned daily, most of that process won't be required.

As the resolution becomes higher and millions of scans are compared false positives will reduce dramatically.

We're going to find huge numbers of benign irregularities we didn't know of.