Is radiography safe from automation?

[mad_scientist]

I'm not talking about radiologists (which I definitely know is not safe from automation) but radiographers.

Radiographers are the people who help position the patient onto the scanner and make them feel comfortable and take the images of the patients' inside body using ct/MRI machines etc.

What does everyone here think?

[fiveworlds]

I could automate it no problem.

[mad_scientist]

32 minutes ago, fiveworlds said:

I could automate it no problem.

This is a serious question, so please only serious replies...

[Klaynos]

2 hours ago, mad_scientist said:

This is a serious question, so please only serious replies...

I think he was being serious. A large part of what they do could be automated. But the putting the patient at ease is harder. But then you don't need to be as qualified to do that so I think his answer stands. 

[Prometheus]

Let's think about what a radiographer actually does.

There's arranging for patients to be picked up from wherever and delivered to radiology: if that could be integrated with the tech nurses use to track patients then this would make things much more efficient.

Then there's positioning the patient appropriately for the scan. Easy for mobile patients, but remember if you're having a scan there is likely something wrong with you so mobility may well be limited. In the worst case you will have an ITU patient - unconscious, critically unwell, supported ventilation and various infusions (and all the equipment that comes with this). Most patients would be somewhere between these two extremes, but i suspect this manual handling is probably the hardest part for automation. It is also vital for achieving optimal imaging.

Then they push some buttons to take a picture. I'm sure this could also be automated quite easily - although many parts have not yet been, though i'm sure the tech exists.

There are also those times someone wants to take an image in some new and weird way. Explaining it to a radiographer is quite easy. Getting a computer to do something outside its code would require bringing in a programmer, getting the programmer to understand the problem and execute a solution. How troublesome this all is depends on the availability of the programmer.

 

At the moment i suspect it is a lot cheaper to employ radiographers than to automate the process. Not sure how long that will last for though. There will also be (already is) resistance from healthcare professionals, who will use unions, lobby government and stoke the populations' fear of robots to maintain the status quo for as long as possible. They just need get enough legislation in place, or people refusing robot services, to make automation not cost effective.

I could well imagine some people paying extra for healthcare with humans after automation in the healthcare industry really sets in, maybe even becoming a status symbol (I can afford human healthcare don't you know). 

 

6 hours ago, fiveworlds said:

I could automate it no problem.

Appreciate the enthusiasm, but you sound like a computer scientist who has never spent a day on a hospital ward.

[Endy0816]

Aren't there new less surrounding scanners coming out? Thought I read something.

Robot skin would help solve the vice grip issue. You can also have semi-automation setups. A human on call who remotely assumes control. 

[EdEarl]

Many jobs, this one included, can be automated in steps. Hospitals often move patients who are confined to bed, and it may take up to four people to do that, once to leave their room, once to position the person in or on the radiology table, then to put them back on a gurney, and finally to put them back in bed. There are lifts to help with it, but they are inconvenient. I think the job of moving patients is a candidate for automation. One person could manage the robot that moves people, saving hospitals considerable manpower. Other more technical jobs will fall to automation eventually.

While robots are getting quite capable, they are still being made of metal, which can be dangerous around people. Thus, things AI can do without a robot will be done by automation before things that need robots around people, in general. Doctors jobs can be done by AI already. Cancer diagnosis, eye diagnosis, and reading radiology scans are things AI can already do better than doctors. This kind of job will fall to automation before medical robots are common.

https://www.newyorker.com/magazine/2017/04/03/ai-versus-md

[Prometheus]

1 hour ago, EdEarl said:

I think the job of moving patients is a candidate for automation.

I don't think so. I'm sure robots could be built to perform this function, but the question is whether it will be more cost effective than hiring humans.

Let's take a simple example.

Bed bound patients at risk of pressure sores are turned every so often - every 2 hours is typical (in addition to having pressure relieving matresses). We need to design a machine that can recognise the position a patient is currently in then somehow physically move the patient onto the other side without hurting or scaring them (people requiring turning are often confused) all the while avoiding pulling out various catheters which may be insitu - a high precision machine would be required with many fail safes. Would it be one machine per bed or per ward? How often would it break down and require humans do to the job anyway? Can the machine perform any other tasks other than turning or would we require another machine to put patients on bed pans? How would you ensure infection control procedures are followed after each episode, would it self-clean (if so, how?) or would you have to employ a human to clean it? Can it leave the ward should the patient have to go elsewhere for a period?

Can anyone come up with an estimate of capital and maintenance costs for such a machine?

Compare this to paying someone near minimum wage who can perform all these asks, requires only limited training and can be extremely flexible in terms of performing multiple manual handling tasks and can move off the ward if need be (including chasing that surprisingly spritely old man down into the car park who keeps trying to escape because he is confused and thinks he's in prison, but AI doctor thinks it too risky to give any sedatives). 

I think it will be a long time until such manual work can be provided by robots cheaper than can be provided by humans (though this also depends on union action level of minimum wage etc). Far easier to replace a doctor than a nursing assistant. 

 

[EdEarl]

I agree that a person should prepare the patient for lift. I'm suggesting the awkward crane like devices they use now can be improved using robotics. Instead of having a person pull it into position for the lift, a robot crane could walk to the patient, position itself, attach to the lifting sheet, lift the patient, and place them on a gurney. The cost of such a device should be under $50K, which is about one salary 24x7x365. A bargain.

[Prometheus]

But you'd still need someone to place the lifting sheet under the patient - they can't be left there for numerous reasons. That's the hardest bit. You'd still need to employ someone to this. But maybe i misunderstand as we basically have machines that can do this now so did you mean something else?

[fiveworlds]

Quote

I Appreciate the enthusiasm, but you sound like a computer scientist who has never spent a day on a hospital ward.

I'm not being enthusiastic the radiographer rarely moves the patient most of the time the nurses do that. All that is involved to replace them is making an easy to use hydraulic based platform with weight sensors and an automated camera that does what it is told. All the nurse would have to do is move the patient to the platform and push a button on the wall (maybe with an id scanner too).

 

 

Edited September 6, 2017 by fiveworlds

[Prometheus]

34 minutes ago, fiveworlds said:

I'm not being enthusiastic the radiographer rarely moves the patient most of the time the nurses do that.

Nurses and radiographers work together to move patients. Even porters get involved sometimes (though technically they're not supposed to). Some patients require no people to help move them, other patients need at least four, but six would be handy. There is also the fine movements radiographers do to get a patient in just the right place based on exactly what image needs to be taken and why.

 

36 minutes ago, fiveworlds said:

 All that is involved to replace them is making an easy to use hydraulic based platform with weight sensors and an automated camera that does what it is told. 

If it's that easy please explain exactly how this would be done. I suspect there are many things you have taken into account but without details i just don't know. Have you seen many patient transfers in radiology?

I'd recommend auditory feedback too for when a patient yelps when something unexpectedly hurts them.

While i remember the machine will also need either to be able to safely log roll an immobilised patient so they can safely vomit (or maybe a suctioning system). Happens only sometimes but can be fatal when it does.

 

[iNow]

We could also rebuilt the machine in such a way that the bed adjusts to the patient, not the other way around. Algorithms can help with that.

[fiveworlds]

Quote

If it's that easy please explain exactly how this would be done. I suspect there are many things you have taken into account but without details i just don't know. Have you seen many patient transfers in radiology?

Yeah I did supernumerary placement in a hospital years ago before I found out that I was immune to certain vaccinations and am not allowed to do that anymore. There are many details to be taken into account personally I have worked with orthopaedics and have brought patients to the mri, ct scan, xray etc. 

Quote

I'd recommend auditory feedback too for when a patient yelps when something unexpectedly hurts them.

That is a great idea and is certainly possible with today's technology. There are apps in development which can do things like automatically translate spoken english into another language and back.

Quote

While i remember the machine will also need either to be able to safely log roll an immobilised patient so they can safely vomit (or maybe a suctioning system). Happens only sometimes but can be fatal when it does.

Absolutely but in such cases maybe it would be better if the nurse stayed in the room like they usually do. Certainly putting a suction system in is easy enough.

To implement such a system properly it would require a large database of patients and would have to know if the patient has certain conditions to do a specific thing. What you would have is a basic system that does nothing at the start just records exactly what the radiographer would do in a certain situation for patients with specific illnesses.  Then over the course of a number of years the data from many such systems would be merged into a working system that should be able to copy exactly what the radiographers did.