Seems like pathology is a white collar job that could be replaced by AI. But the pathologists I know are not worried about their jobs, (honestly, they are worried about it, but for other reasons - mostly changes in the US healthcare system).
Its hard to imagine no human in the loop. My guess is that in the future, there might be fewer pathologists, or may be the same number, just spending time doing different things (perhaps reviewing results for tests we haven't heard of yet) or just validating what the machine thinks. This being one of the tools that makes them more productive.
Between the growth of medical technology and the aging population I don't think they will every worry about finding work.
Indeed, I do not think either that doctors need to worry much. AI will change our jobs for the better, that is pretty sure. Maybe we will earn a little less, but what of it?! Putting us out of work would mean handling the huge amount of technical details involved in our daily jobs. People (and HN is no exception to that) really, really underestimate that. The biggest problem is the amount of rigorous feedback you need to train an efficient system. The medical system is a long, long way from having such information retrieval capabilities. And then you have the physical procedures... Robots are also still have a long way to go on that one!
Unsurprisingly perhaps I'm not so sure. Could you elaborate on the processes you allude to? I'd be very interested.
A pathologist I know attends cancer surgeries and analyses biopsies while the surgeon waits to make a decision. Very little process there, but a fair amount of time and hence cost and risk. A robot here might improve outcomes and costs, and seems reasonably realistic. And the surgeon is not immune from the threat to jobs posed by robots.
Another role I can imagine in pathology might be for a human to randomly spot-check diagnoses made by image analysis. That would become comparatively very mundane and potentially low paid. And since humans tend to perform worse than AI in some image analysis tasks, I can envisage a time where the robots are trusted 100% and the humans are seen as anachronistic.
In summary, I guess, complex processes might not protect jobs, but make them a clear target for efficiency improvements.
I am not saying medicine entails complex processes in terms of intellectual capabilities. Quite the opposite, actually. The complexity I was speaking about is that of actual physical procedures. Most specialties span a much larger range of activities than one imagines at first glance. For example, pathology does involve looking at surgical samples under the microscope, but that is only the final step in the process. You first have to describe the specimen macroscopically, and make sense of its original location and orientation in the body. Autopsy, a process akin to surgery, also falls in the realm of pathology.
Granted, some specialties will be easier than others to automate. But the complexity of the machine feedback processes that will have to be implemented are currently overwhelming in most specialties. We will see a long period during which AI facilitates our work, without replacing us, due to lack of integration in the workplace.
You mentioned surgery as difficult for AI, but interestingly it is one of the fields where feedback will be easier to implement, since patients rarely undergo surgery without extensive imaging first. But in many technical procedures (anesthesia, intubation, catheterization, bronchoscopy, etc...) you often perform without fixed patterns, and orient yourself using the (quite variable) patient environment and symptoms.
There is no question this all will be automated someday. But it is far more complex than working on a car assembly line. We are not there yet.
Thanks for the reply. I didn't known that autopsy was a pathology discipline.
Personally I suspect that many medical roles will be broken into sub-roles and semi automated. Humans will remain with stop/go decisions and overview, for a while at least. I think we are probably in agreement.
Incidentally, I agree regarding surgery, perhaps I was unclear.
Indeed, I spoke with a pathologist friend about this. Non-experts assume its black and white (cancer or not) but there are many many shades of grey and other variables. Its very complex.
Between the growth of medical technology and the aging population I don't think they will every worry about finding work.
Frankly, I think there are many medical specialties where doctors do need to worry about finding work.
E.g. go back 13 years and this was already happening: Enter the “nighthawks,” a generic moniker applied to groups of American-trained diagnostic radiologists located in India and Australia who provide immediate diagnostic interpretation of CT imageshttp://www.hopkinsmedicine.org/about/Crossroads/09_09_04.htm...
That's only going to get worse. There are many jobs that can be shipped overseas.
I really think that in the short term when these ML techniques are available (and proven since this appears to be unpublished with no replication / peer review) are going to be tools for Physicians.
Also, it might be that the role of doctors changes. Maybe you have an automated system that will do all the medical testing but then refer you to someone that can create a treatment plan , perform operations and so forth.
Treatment plans are far more complex than diagnosis. Much of it is convincing the patient to make the right choice. An AI that can sympathize and educate is a long time coming.
From what I've heard, patients are often more aware of their own bodies than people give them credit for. Ie. If someone thinks they're dying, you should believe them.
Having another human there to reassure a scared patient that a treatment plan gives them their best chance is invaluable.
Also I'd be worried about pharmaceutical companies having an influence on the AI - as in making sure it doesn't know about generics and having it aggressively push their new drugs for any applicable cases. A human would be much more resilient to such machinations, on average.
Additionally, trying to elicit the information you need to make a diagnosis is much harder than non-clinicians realise. Present a series of yes no questions doesn't really capture all of the information that is required; I suspect this is partly why previous expert systems have failed thus far.
From the moment the patient walks into the room you are assessing their clothing, their colour, their gait, their physique, how they're breathing, their emotional state, their physical abilities.
When you are discussing their problem you are varying not only the content of the questions you ask but also the style and the wording to match the patient's level of education, understanding, local culture and terminology etc.
For certain conditions patient's will lead you completely astray if you take what they are saying as canon. There is an art to human conversation and the best doctors I know excel at this as well as having the depth and range of knowledge to synthesise a diagnosis.
After a history you will typically examine the patient. Listen to their heart and lungs, palpate their abdomen, look at their retina or their eardrum, etc.
The diagnosis is made by gathering all of this information together and then running a "pattern matching routine".
I have no doubt that machine learning software is coming for us, but the job that doctors too is often oversimplified by techies. Myself included. As a programmer and electronic engineer before starting medicine I thought that it would be trivial to replicate much of the work in software. It isn't. It will happen but after another couple of orders of magnitude improvement.
In the meantime, using machine learning for diagnostic aids is hugely valuable. The sort of things we are seeing at the moment are very much the "low hanging fruit" of medicine - that is, tasks for which the input, analysis and output is well defined and easy to feed into a computer algorithm. Examples include analysing histology specimens for cancer, analysing head CT scans for stroke, analysing cardiac monitoring for patients at high risk of heart attack, and so on. The "difficult" part is the data gathering.
As an aside, I do think the role of doctors has changed over time. Medicine has advanced enormously as a field over the last hundred years. Much of what used to be the role of the doctor is now the role of nurses, nurse specialists, nurse practitioners or even physician assistants. Yet still (in the UK at least) doctors have a never ending amount of work to do!
I'm very excited to be at the crossroads of medicine and technology and upbeat about what the future will bring for patient care.
Related to this vaguely, I wonder how easy it would be to code a cell counting app for a phone.
Awhile back, I designed and printed a case for my phone that could fit over the lens of a microscope allowing me to take pictures/videos with it. One technique that researchers use to estimate the number of cells in a dish/suspension is to aliquot a small amount and place it on a hemocytometer. Essentially, this is a slide with etched grids; counting the number of cells in the area allows you to estimate the larger population. It is tedious and depending on the number of cells, easy to mess up.
I'd imagine that phones now are powerful enough to analyze this. I don't know whether a live-imaging setup is capable as I am no expert in phone tech. But taking a picture, thresholding to remove noise, etc... seems like it could be done.
I was first author on the pigeon paper referenced below. It was a lot of fun to do, but afterwards, we found that a convolutional neural net with transfer learning could outperforscale interaction with them at the UCDH level. Already there is interest in working with Chris Polage on his C. difficile activities.
m them, at least on mammogram images.
Its hard to imagine no human in the loop. My guess is that in the future, there might be fewer pathologists, or may be the same number, just spending time doing different things (perhaps reviewing results for tests we haven't heard of yet) or just validating what the machine thinks. This being one of the tools that makes them more productive.
Between the growth of medical technology and the aging population I don't think they will every worry about finding work.